Molecular Imaging in Cancer: Linking Biology,Function and Clinical Application In Vivo

ABSTRACTS

Abstract ID: 386

Aposense—A New Tool for In Vivo Detection of Chemotherapy-Induced Tumor Cell Death

Anat Shirvan, Galit Levin, Tali Aloya, Maya Damianovich, Amnon Farkas, Merav Bassan, Ilan Ziv

NST Ltd. Petach-Tikva, Israel, 49170

A common current view in oncology is that anticancer treatments exert their effect by induction of apoptosis in their target tumor cells. The sensitivity to cell death is therefore a major determinant in the success of anticancer therapy. However, the response of tumors to therapy suffers from lack of measures that can rapidly and noninvasively assess the effectiveness of the drugs. Monitoring of chemotherapy is currently based mainly on gross evaluation of tumor size changes, after weeks or months of treatment, during which the patient suffers from severe side effects and does not always benefit from the chemotherapy. Therefore, fast determination of the cell death response of tumors to anticancer drugs is of utmost importance. Aposense (NST820, IL patent 141571) is a small molecule detector designed to recognize apoptotic cells early upon apoptosis induction. Subsequently, the Aposense molecule selectively binds, penetrates into, and accumulates within apoptotic cells. We have successfully used a fluorescent derivative of the Aposense molecule to follow the response of tumor cells to chemotherapy treatment in vivo, and to monitor untoward adverse effects of the chemotherapy in healthy tissue, namely, apoptosis of gastrointestinal epithelial cells in the crypts of the small intestine. In an animal model of mice bearing subcutaneous c-26 tumors, the effect of the newly developed slow release chemotherapeutic drug, doxyl, was assessed after intravenous injection of Aposense. At different time points following chemotherapy, mice were sacrificed and histological sections were evaluated. The kinetic response to the chemotherapy (viewed as increased uptake of Aposense into apoptotic cells) exhibited two waves of cell death following one dose of doxyl treatment, characterized both by an increase in the number of identified apoptotic events and by an increase in the fluorescent intensity of intracellular-bound Aposense. The use of another chemotherapy protocol (a combination of Taxol and cyclophos-phamide) revealed a different kinetics of cell death response. A strong correlation was observed between the number of tumor cells that were identified by Aposense after in vivo administration and those that were ex vivo identified by TUNEL assay. Due to the nonselectivity of the chemotherapy treatment, extensive apoptosis was observed in the epithelial cells in the crypts of the small intestine. The use of doxyl chemotherapy protocol triggered two peaks of extensive apoptotic process in the small intestine 60 and 100 hr following treatment. The kinetics of the response changed when Taxol and cyclophosphamide were used, culminating in one peak of cell death 24 hr past treatment. In summary, the Aposense has been shown as a sensitive and fast tool that can provide optimization of anticancer treatments by both in vivo monitoring the effectiveness of different anticancer drugs on target tumor cells and assessment of side effects in healthy tissues. Future radiolabeling of Aposense will therefore turn it into an efficient and noninvasive measure that could assist the physician in imaging the response of cancer cells to chemotherapy, significantly improving anticancer protocols.

Abstract ID: 407

Monitoring the Temporal Distribution of Taxol^® in a Living Tumor Cell Using Raman Imaging Microscopy

Jian Ling, Michael A. Miller, Edna Cruz, Steven D. Weitman, Alan C. Bovik

Direct Raman imaging techniques are being developed to study drug distribution in living cells. The advantage of Raman imaging over the fluorescent imaging is that no external markers are required, which makes sample preparation much simpler for experiments. At the same time, the mechanism of action of the drug is minimally disturbed during imaging. In this study, Raman imaging was used to monitor the change of Taxol^® (paclitaxel) distribution in cultured human breast tumor cells. Approximately 10⁵ MDA-435 breast tumor cells were cultured on gold-coated Petri dishes and were allowed to stabilize for 24 hr in an RPMI-1640 medium supplemented with fetal bovine serum. Before imaging, the RPMI nutrition medium was exchanged with a phosphate-buffered salt (PBS) solution. Control images were first obtained while the cells were in the PBS solution. Cells were then treated with Taxol (0.3 mg/ml or 350 μM) for 1 hr. After drug treatment, the Taxol solution was washed out and the cells were put back into the PBS solution. Images of the same cell were acquired before, during, and after the Taxol treatment. Each measurement is composed of three images of the cell: a conventional white-light image and two Raman images acquired at the 1000 and 1080 cm⁻¹ Raman bands, respectively. The white-light image is used to record the cell structure and any changes that might occur to it. The Raman image acquired at the 1000-cm⁻¹ records Taxol distribution in the cell as well as background fluorescence from the cell. Using the Raman image acquired at 1080 cm⁻¹, the contribution of background fluorescence present in the 1000-cm⁻¹ image can be eliminated, thereby leaving behind only Taxol-related information. The superimposition of the Raman and white-light images illustrates how the distribution of Taxol changes with time in live tumor cells. It was found that Taxol does not enter the cell nucleus under these conditions, but is more concentrated around the cell centrosome and near the cell membrane. This result is well explained by the accepted binding characteristics of Taxol and its molecular target—the microtubules. In conclusion, this work demonstrated the feasibility of using direct Raman imaging to study the distribution of anticancer agents in single living cells. Based on this study, direct Raman imaging can be further extended to study drug mechanisms, cellular uptake, resistance, and intracellular pharmacokinetics.

Abstract ID: 409

Gene Expression Fingerprints for Molecular Classification of Cancer

Ivana V. Yang¹, Emily Chen¹, Jeremy P. Hasseman¹, Domenico Coppola², Timothy J. Yeatman², John Quackenbush¹

¹The Institute for Genomic Research, Rockville, MD and ²H. Lee Moffit Cancer Center, University of South Florida, Tampa, FL

Molecular classification of cancer promises to lead to cancer prevention, early diagnosis, and the development of optimal treatment protocols. Histologically, similar tumors may result from substantially different genetic changes; the ability to capture these underlying genetic changes will be essential for the development of new tools. Furthermore, molecular fingerprints may prove critical in diagnosis of clinical cases where traditional methods have failed; one example is “unknown primary” cases where a metastatic lesion is detected but the primary site of cancer is not known. The first and crucial step in molecular classification of cancer is the establishment of tissue-specific expression patterns. We have completed expression profiling of adenocarcinomas from eight different organs (breast, colon, esophageal junction, kidney, lung, ovary, pancreas, and stomach) using 32,448-element cDNA microarrays. Hybridizations were performed in duplicate on 10 patient samples representing each class, and tissue-specific sets of genes extracted from the data set using a classifying algorithm. The fingerprints identified in this study will be used in the future for classifying unknown primary metastases.

Abstract ID: 411

Functional Studies of CHK and Csk in Breast Cancer

Cécile Bougeret¹, Shuxian Jiang¹, 1Iafa Keydar², Hava Avraham¹

¹Division of Experimental Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA 02115, USA and ²Department of Cell Research and Immunology, Tel Aviv University, Ramat Aviv 69978, Israel

In this report, we analyzed the expression and kinase activities of Csk and CHK kinases in normal breast tissues and breast tumors, and their involvement in heregulin (HRG)-mediated signaling in breast cancer cells. Csk expression and kinase activity were abundant in normal human breast tissues, breast carcinomas, and breast cancer cell lines, while CHK expression was negative in normal breast tissues and low in some breast tumors and in the MCF-7 breast cancer cell line. CHK kinase activity was not detected in human breast carcinoma tissues (12/12) or in the MCF-7 breast cancer cell line (due to the low level of CHK protein expression), but was significantly induced upon HRG stimulation. We have previously shown that CHK associates with the ErbB-2/neu receptor upon HRG stimulation via its SH2 domain and that it down-regulates the ErbB-2/neu-activated Src kinases. Our new findings demonstrate that Csk has no effect on ErbB-2/neu-activated Src kinases upon HRG treatment and that its kinase activity is not modulated by HRG. We further characterized CHK function in HRG-mediated signaling and evaluated the antitumoral potential of overexpressing wild-type (wt) CHK protein in MCF-7 cells by stable transfection. CHK significantly inhibited in vitro cell growth, transformation, and invasion induced upon HRG stimulation. In addition, tumor growth of wt CHK-transfected MCF-7 cells was significantly inhibited in nude mice. Furthermore, CHK down-regulated c-Src and Lyn protein expression and kinase activity, and the entry into mitosis was delayed in the wt CHK-transfected MCF-7 cells upon HRG treatment. These results indicate that CHK, but not Csk, is involved in HRG-mediated signaling pathways, down-regulates ErbB-2/neu-activated Src kinases, and inhibits invasion and transformation of breast cancer cells upon HRG stimulation. These findings strongly suggest that CHK is a novel negative growth regulator of HRG-mediated ErbB-2/neu and Src family kinase signaling pathways in breast cancer cells.

Abstract ID: 417

Molecular Imaging of a Murine Melanoma Model Using a Novel Targeted Blood Pool Agent That Binds Integrins

Samira Guccione, Dan Lee, Mark Bednarski

Department of Radiology, Stanford University, Palo Alto, CA 94305

We have developed a molecular imaging agent that targets a class of cell surface receptors associated with tumor angiogenesis, the integrins, in an effort to spatially resolve and map their expression in vivo. The integrins α_vβ₃ and α_vβ₅ have been identified as important cell surface markers of angiogenesis found to be overexpressed in the majority of solid malignant tumors. We have used a nanoscale lipid-polymer-based agent (NP, described elsewhere) with a multivalent array of a novel synthetic peptidomimetic (PM) to specifically target this class of receptors known to be expressed on vascular endothelial cells and certain tumor cells. By labeling the PV–NP with ¹¹¹Indium, a gamma-emitting radioisotope, we were able to obtain whole-body scintigraphic images in a murine model of malignant melanoma. Tumors were implanted into test animals, grown to significant mass, and then scanned following the administration of a single bolus intravenous injection of ¹¹¹Indium-labeled PM–NP. Time course scanning demonstrated a persistent blood-pooling effect, and accumulation of the PM–NP at the tumor. Tumor accumulation was dependent on the presence of the peptidomimetic moiety since nontargeted contrast agent did not show significant tumor localization. Furthermore, integrin specificity of the PM was examined both in vitro using a cell-adhesion blocking assay and in vivo by immunohistochemistry and autoradiography. We have found that the PM–NP was functionally superior to the monomeric peptidomimetic in our in vitro assay, demonstrating the increased binding avidity conferred by a multivalency effect. We have also observed this increased binding avidity in vivo as a function of percent PM moiety on the NP. This approach to visualizing receptors associated with clinically relevant pathological processes has a general application. We propose to use this type of molecular imaging agent to confirm and to follow the progression of tumor angiogenesis during cancer therapy using clinical medical imaging techniques.

Abstract ID: 418

Molecular Imaging of Platelet-Derived Growth Factor Receptor Using Magnetic Resonance Image-Guided Functional Genomic Analysis as a Method for Target Identification

Samira Guccione, Carina Mari, Yi-Shan Yang, Dan Lee, Mark Bednarski

Department of Radiology, Stanford University, Palo Alto, CA 94305

The recent explosion of information in the fields of genomic and proteomic has provided a rich ground for the discovery of molecular targets for therapeutic and diagnostic purposes. However, due to the heterogeneity within most tumors, tissue sampling for genomic and proteomic analysis is spatially dependent. In addition, molecular targets vary with temporal changes during tumor growth. We believe that there is a need to resolve tissue analysis both spatially and temporally so that the most relevant molecular targets can be identified. Clinically, regions of high contrast enhancement in magnetic resonance imaging (MRI) are believed to be related to areas of increased vascularization, angiogenesis, and tumor aggressiveness. We hypothesized that MRI-guided tissue sampling from the contrast-enhanced (CE) and nonenhanced (NE) regions within the same tumor will facilitate the identification of molecular targets. To test this hypothesis, a murine squamous cell carcinoma model was used to obtain tissues from CE and NE areas of the tumor after MRI. While histologic staining (H&E) showed no significant differences between the CE and NE regions, the MR images clearly showed differences in the perfusion and diffusion of the small molecule contrast agent (Gd-DTPA). Oligonucleotide microarray analysis of these spatially distinct areas within the same tumor revealed that nine genes were up-regulated in the CE areas of the tumor relative to the NE regions. Only one gene was down-regulated. Six out of these nine up-regulated genes in the CE regions are associated with the extracellular matrix (ECM). The only major growth factor gene up-regulated in the CE areas was the platelet derived growth factor receptor (PDGF-R). Both PDGF and its receptors have been shown to have an angiogenic effect. Furthermore, it has been proposed that the overproduction of PDGF may be involved in autocrine and paracrine growth stimulation of human tumors. Activation of this receptor may be of critical importance in tumor progression. Therefore, PDGF-R was identified as a potential target of interest. The distribution pattern of this receptor was evaluated using immunohistochemical staining (IH). IH receptor distribution matched the MRI contrast pattern, confirming a positive target identification. Molecular imaging of this target was performed using whole body nuclear scintigraphy with ¹²⁵I-labeled anti-PDRG-R antibody alpha. Results were further confirmed by autoradiography, and blocking experiments with noniodinated antibody. There is a remarkable correlation between the spatial distribution of PDGF-R and the corresponding contract enhancement pattern of the tumor MR images. We have demonstrated that MR-guided genomic analysis of spatially heterogeneous tumor samples can aid in identification of target selection. In addition, these data imply a potential role for PDGF-R in angiogenesis and metastasis in the CE regions of the tumor at this stage of tumor growth. Further experiments will help understand the biological role of PDGF-R in the CE tumor area.

Abstract ID: 419

Parametric MRI of Breast Cancer Progression and Metastasis in Animal Model

Maya Chetrit-Dadiani, Raanan Margalit, Dalia Seger, Ori Brener, Hadassa Degani

Department of Biological Regulation, Weizmann Institute of Science, Rehovot, Israel

Imaging of breast cancer metastasis in animal models is a challenge due to the small size and random distribution of the metastases. We applied high-resolution (in-plane 0.08 mm²) dynamic contrast-enhanced MRI to characterize the histopathology features and angiogenic capacity of primary tumors and metastatic spread in a breast cancer animal model. Human breast cancer cells (10⁶ – 10⁷) MDA-MB-231 were orthotopically implanted in the mammary glands of SCID mice. Metastases were detected in hematoxilin–eosin stained sections of the lungs, bones, and lymph nodes 32–46 days after implantation. The MR experiments were performed on a 4.7-T Biospec spectrometer (Bruker Analytic, Germany). High-resolution T2-weighted images and 3-D gradient-echo dynamic contrast-enhanced MRI (Gd-DTPA) were applied to monitor tumor progression. A model-based analysis of the contrast-enhanced images was applied to the primary tumor and to the entire region of the hindlimb bones. The analysis yielded parametric color-coded images of vascular permeability × surface area (PS) and extracellular volume fraction (EVF) accessible to the contrast agent (Furman-Haran et al. J Magn Reson 1997;128:161). In addition, two color-coded maps, maximum enhancement as well as time to maximum, served to identify the metastatic loci. The MR images and parametric maps were compared to histology findings. The volume and histopathological features of the primary tumor were determined by analyzing the T2-weighted images. In most of the tumors, a central cyst-like region was observed, suggesting insufficient drainage. Following tumor growth, the fluid in the cyst accumulated cell debris and blood remains as was indicated by the darkening of the signal in this region. A delayed enhancement was observed in the cysts due to slow diffusion from surrounding vascularized regions. The parametric images of the PS and EVF distribution in the primary tumor demonstrated high inter- and intratumoral heterogeneity with PS ranging from 0 to 0.2 min⁻¹ (mean 0.023 ± 0.025 min⁻¹) and EVF from 0 to 1 (mean 0.204 ± 0.15). Metastasis localization was first determined from the histology sections of the bones. The metastatic loci in the MRI images were then resolved in the maximal enhancement maps. Further analysis using the model-based approach indicated similar PS and EVF values in the identified metastatic loci to those of the primary tumor, suggesting high similarity between the vascularity of the primary and metastatic lesions. In conclusion, we have shown the importance of imaging the inner compartmentation of tumors (viable regions, necrosis, or cyst-like regions) in order to accurately monitor changes in tumor progression. We also demonstrated the potential capacity of high-resolution contrast-enhanced MRI to detect small loci of metastasis in a breast cancer animal model. Specific values of the vascular parameters found in the primary tumor may serve to identify metastatic loci with similar vascularity. Further studies to investigate tumor progression, invasion, and metastasis in this tumor model, using fluorescent markers, are currently underway.

Abstract ID: 420

Image-Guided Proteomics in Human Gliomas

S. K. Hobbs¹, G Shi¹, R Homer², G Harsh³, S. W. Atlas², M. D. Bednarski¹

¹Lucas MRS Center, ²Department of Radiology, and ³Neurosurgery, Stanford Medical Center, Palo Alto, CA 94304

The focus of our work is to develop new methods to use imaging techniques to guide tissue sample selection for genomic and proteomic analysis. Here we describe the use of image-guided protein expression profiling on three malignant human gliomas: a primary and recurrent glioblastoma multiforme (GBM), and primary gliosarcoma. Gliomas are the most common type of brain cancer and GBM is their most malignant variety. Average survival for patients with GBM is approximately 12 months. The classic MRI appearance of GBM reflects the tumor's heterogeneous pattern of hypercellularity, necrosis, neovascularity, and invasion. Recent microarray analyses have shown both heterogeneity of gene expression within the same tumor as well as among different GBM tumors. Since these previous studies lack image guidance in sample acquisition, we began to investigate image correlates to protein expression patterns in gliomas. In this study, Gd-contrast-enhanced MRI was used to direct tumor specimen harvesting. Separate samples from contrast-enhancing (CE) and nonenhancing (NE) regions were flash-frozen and analyzed for protein expression using proteomic Matrix Associated Laser Desorption/Ionization-Mass Spectroscopy (MALDI-MS). This analysis performed on modified MALDI surfaces showed that in general, protein expression profiles of CE regions differed from those of NE regions. Among the differences was a 10-kDa protein peak seen primarily in the CE regions which likely represents matrix Gla protein (MGP), a protein whose mRNA expression has been shown to be up-regulated in tumors. By combining sophisticated imaging tools such as MRI, stereotactically selective tumor sampling, and new proteomics technology, specific patterns of protein expression can be identified in gliomas. This identification of diagnostic markers and therapeutic targets within gliomas holds promise for improved management of the disease through the development of molecular imaging agents for better diagnosis, pathologic feature correlates, and follow therapeutic effectiveness.

Abstract ID: 429

PAI-1 Promoter-driven PAI-1-GFP Expression Vectors for Direct Assessment of Transcriptional Activity in Individual Cells In Vivo

Paul J. Higgins, Jianzhong Tang

Center for Cell Biology and Cancer Research, Albany Medical College, Albany, NY 12208

Studies in plasminogen activator inhibitor type-1 (PAI-1) null mice have disclosed a requirement for this SERPIN in tumor invasion and angiogenesis in vivo. PAI-1 is the major physiologic regulator of urokinase plasminogen activator and, thereby, of plasmin generation. Plasmin initiates a proteolytic cascade utilized by both tumor cells and activated endothelial cells to dissect and locomote through extracellular matrix/stromal barriers during tissue invasion. PAI-1 tissue levels and synthesis must be carefully controlled in this process as deficient PAI-1 expression (as in the PAI-1 null mouse) or excessive PAI-1 production is inconsistent with maintenance of a matrix scaffold permissive for tumor invasion or formation of functional neocapillary structures. PAI-1 has emerged, therefore, as a critical regulator of the pericellular proteolytic balance required for both metastatic spread and a successful tumor-associated angiogenic response. This SERPIN affects cellular motile events directly, by controlling the extent and localization of matrix barrier proteolysis, as well as by influencing cell-to-matrix adhesive properties. To evaluate relationships between PAI-1 expression and cellular migratory activity, 800 bp of the human PAI-1 promoter were cloned upstream of a GFP insert and this expression reporter construct transfected into keratinocytes and microvessel endothelial cells. PAI-1 transcription was confirmed by Northern blotting (to evaluate motile effects on the endogenous PAI-1 gene) and, at the single cell level in vivo, by measurements of GFP reporter activity after the switch from a sessile (i.e., contact inhibited) to an actively locomoting phenotype. Plating of PAI-1 promoter-GFP reporter transfectants at low density to stimulate cell movement resulted in a >30-fold increase in GFP expression and a similar stimulation in the rate of transcription of the endogenous PAI-1 gene. GFP expression correlated with the real time of induced cell motility, was activated only in the motile cohort, and was suppressed upon cessation of cell locomotion, indicating that this vector system provided a tool to visualize PAI-1 transcriptional activity during in vivo migration. To assess this more directly, a second vector was developed in which 800 bp of the human PAI-1 promoter were cloned upstream of an insert encoding a PAI-1-GFP chimeric protein. De novo synthesized PAI-1-GFP was tracked in situ by fluorescence microscopy and followed the normal route of Golgi accumulation, cytoplasmic packaging, and vitronectin-dependent matrix deposition. Such PAI-1–expressing cells were highly motile as quantified in a 2-D planar model of basal and growth factor-induced migration. Quantitation of the level of chimeric protein expressed, furthermore, correlated with the relative migratory activity of both keratinocytes and endothelial cells. PAI-1, moreover, was a critical element in stimulated cell movement since attenuation of PAI-1 synthesis by constitutive expression of PAI-1 antisense vectors ablated the migratory response. These findings illustrate the usefulness of this approach to image transcriptional and behavioral responses in single living cells using promoter sequences from genes implicated in cellular invasive traits. [This work was supported by grants from the NIH (GM57242) and the Department of the Army (DAMD17-98-1-8015 and DAMD17-00-1-0124).]

Abstract ID: 440

MRI Monitoring of Prostate Tumor Growth in Transgenic Mice

A. Elgavish^1,2, J. Wilbanks², P. Kiss³, L. Lenard³, M. Schuler¹, R. Mentor-Marcel¹, T. Cartee¹, L. Tian¹, A. Wood¹, C. A. Pinkert¹, G. A. Elgavish³

¹Genomics and Pathobiology, ²Urology and Biochemistry, and ³Molecular Genetics, University of Alabama at Birmingham, Birmingham, AL, 35294-0019, USA

Objective: p53 mutations are often found in advanced prostate cancer. To test the hypothesis that p53 mutations facilitate progression of prostate cancer, we generated a transgenic mouse model with human mutant p53 expression in the prostate.

Methods: A transgenic mouse line was generated with a construct using: (i) a − 426/+ 28 fragment of the rat probasin promoter (rPB) (a gift from Dr. R. Matusik); and (ii) a p53 expression construct containing human mutant p53, resulting in amino acid substitution at amino acid 273 (a gift from Dr. A. J. Levine). Purified rPB-mutant p53 constructs, characterized by extensive restriction and sequence analysis of the ligated junction, were micro-injected into B6SJL F2 mouse zygotes at the UAB Transgenic Animal Facility. Founders were characterized by PCR analysis of tail DNA One germ-line transmitter was bred to establish an individual transgenic line. Expression of the transgene was monitored using tail DNA PCR or Southern analysis. Immunofluorescence microscopy of prostate sections was used to determine expression of human mutant p53 at the protein level. MRI was used to follow the growth of the prostate noninvasively at 18, 24, and 28 weeks of age. MRI was carried out on a Bruker Biospec 4.7, equipped with a custom-made body coil of 50-mm diameter. A multi-slice, multi-echo sequence was used for acquisition. FOV: 8 cm, slice thickness: 2 mm, orientation: transversal, coronal, sagittal; matrix: 256 × 256, recovery time: 2397 msec, number of tomographic slices: at transversal orientation 12, at coronal 4, and at sagittal 8. The volume of the prostate was calculated using the formula V = (4/3) p × r1 × r2 × r3, where r1, r2, r3 were the three radii of the prostate ellipsoid.

Results: Nuclear accumulation of human mutant p53 protein occurred in the prostate epithelium of transgenic mice, but not in nontransgenic littermates. Apoptosis was examined in the prostate epithelium at varying times after castration. The percentage of apoptotic cells was lower in transgenic mice as compared with their nontransgenic littermates. Mice expressing mutant p53 did not develop prostate tumors. The hypothesis that mice expressing an oncogene and mutant p53 in the prostate epithelium might develop prostate cancer at an earlier age was tested in rPB-mutant p53 mice crossed with TRAMP mice. TRAMP, developed by Greenberg et al. (Baylor College of Medicine, Houston, TX), is a transgenic mouse model of prostate cancer with the SV40 small and large T antigen regulated by rPB. The incidence of TRAMP/rPB-mutant p53 mice with enlarged prostate volume at 18 weeks of age was higher than that of TRAMP mice.

Conclusions: Transgenic rPB-mutant p53 mice accumulate mutant human p53 in the prostate, resulting in altered apoptosis in response to castration. Mice expressing an oncogene and mutant p53 (TRAMP/rPB-mutant p53) develop prostate cancer earlier in life.

Abstract ID: 443

Imaging of Apoptosis with In-111 Labeled PEGylated Annexin V

Xiaoxia Wen, Qingping Wu, Peng Huang, Diana Chow

Sidney Wallace, Chusilp Charsangavej, and Chun Li U.T./M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030

Annexin V binds with high affinity to phosphatidylserine (PS) exposed on the surface of apoptotic cells. Tc-99m-labeled annexin V has been proposed for the imaging of apoptosis. Since apoptosis is a dynamic process as newly generated apoptotic cells are constantly removed by phagocytic macrophages, it is hypothesized that prolonged circulating time of radiolabeled annexin V would allow binding of annexin V to newly formed apoptotic cells over a period of time. As a result, increased accumulation of radiolabeled annexin V in tumors and enhanced imaging property is anticipated. Unfortunately, annexin V is rapidly cleared from the blood after intravenous injection. Modification of proteins with polyethylene glycol (PEG) is known to prolong their blood circulation times. To introduce both radiometal chelator and PEG simultaneously to annexin V without excessive modification of the protein, we devised a general approach by introducing metal chelator to one terminus of PEG, which is in turn coupled to annexin V. Thus, heterofunctional PEG with one end containing isothiocyanate (SCN) and the other containing DTPA was synthesized and characterized. Mixing annexin V with SCN-PEG-DTPA in PBS (pH 8.5) at molar ratios of 1:30 or 1:60 yielded PEGylated annexin V. PEGylated annexin V was labeled with In-111 with radiochemical purity >98%. To evaluate the binding of PEGylated annexin V to apoptotic cells, human leukemia HL60 cells and B-cell lymphoma Raji cells were treated with Ara-C at 1.0 mM for 22 hr. A portion of the cells were stained with annexin V-FITC and analyzed by flow cytometry. Another portion was incubated with In-111-labeled PEGylated annexin V and cell-associated radioactivity was measured. Flow cytometry revealed that the percentage of apoptotic cells increased 4- to 6-fold after treatment with Ara-C. Similarly, cell-associated radioactivity was also increased 4- to 6 fold. Interestingly, the 1:60 prep of PEGylated annexin V did not bind to cells treated with Ara-C. These data suggest that the 1:30 prep of PEGylated annexin V binds to apoptotic cells in the same way as annexin V-FITC does, whereas overmodification resulted in loss of binding affinity. To confirm that PEGylation of annexin V does improve the blood half-life of the protein, the 1:30 prep of In-111-labeled, PEGylated annexin V was injected into nude mice and blood was drawn at different times after the injection of the radiotracer. Pharmacokinetic calculation based on a two-compartment model revealed t(1/2, α) of 0.96 hr and t(1/2, β) of 17.1 hr. At 1 hr after injection of the PEGylated annexin V, as much as 37% of injected dose per gram of blood was still circulating in the blood. This is in contrast to reported values of 1.9% for annexin V directly labeled with Tc-99m. Our data suggest that radiolabeled, PEGylated annexin V may be useful for imaging of apoptosis.

Abstract ID: 444

MicroPET Imaging of CEA-Positive Xenografts Using I-124 Labeled Engineered Anti-CEA Antibody Fragments (Diabodies and Minibodies)

G. Sundaresan¹, Paul J. Yazaki², John E. Shively², Khoi N. Nguyen¹, Ron Finn³, Steven M. Larson³, Andrew A. Raubitschek², Sanjiv S. Gambhir¹, Anna M. Wu^1,2

¹Crump Institute for Molecular Imaging University of California School of Medicine, Los Angeles CA 90095, ²Beckman Research Institute and City of Hope National Medical Center, Duarte, CA 91010, and ³Memorial Sloan-Kettering Cancer Center, New York, NY 10021

Despite the ability of intact antibodies to deliver radionuclides to tumor targets in vivo, prolonged clearance kinetics limit their utility for imaging tumors. We previously produced genetically engineered anti-carcinoembryonic antigen (CEA) antibody fragments that display rapid, high-level tumor uptake and rapid clearance from the circulation in the athymic mouse/LS174T xenograft model. The T84.66 anti-CEA diabody (noncovalent dimer of single-chain Fv, 55 kDa) and minibody (scFv-C_H3 dimer, 80 kDa) exhibit features favorable for radioimmunoimaging. The present work evaluates the combination of the minibody or diabody labeled with a positron-emitting radionuclide, I-124 (half-life, 4.18 days), for imaging tumor-bearing mice using a high-resolution small-animal microPET system. Typically, 0.2 mg of protein was labeled using 1.7-2.6 mCi I-124 by the Iodogen method. Labeled protein was purified by size exclusion HPLC. In vivo distribution was evaluated in athymic mice bearing paired LS174T human colon carcinoma (CEA + ve) and C6 rat glioma (CEA − ve) xenografts. Thyroid and stomach uptake of radioiodine were blocked by pretreatment of the mice with KI and KClO₄, 24 hr and 30 min prior to tracer injection, respectively. Mice were injected intravenously with I-124-minibody (63-66 μCi, 8 mice) or I-124-diabody (87 μCi, 4 mice) and imaged at 4 and 18 hr by microPET. Four mice were also imaged using F-18 fluorodeoxyglucose (FDG) prior to I-124 minibody. Immediately after the final scan, the mice were sacrificed and tissue samples were weighed and counted; alternately, mice (n = 2) were subjected to whole-body digital autoradiography. Labeling of the minibody or diabody with I-124, followed by purification, required less than l hr with resultant specific activities of 2.2–3.6 μCi/μg. Radiolabeling efficiencies ranged from from 33% to 88% and immunoreactivity determined by a liquid phase CEA-binding assay was 42% (diabody) or >95% (minibody). MicroPET images showed high uptake of I-124 minibody in CEA-positive tumor (9.51 ± 2.84% injected dose per gram [%ID/g] at 4 hr; 10.7 ± 3.71 %ID/g at 18 hr) compared to control tumor (3.17 ± 1.04 %ID/g at 4 hr; 1.11 ± 0.44 %ID/g at 18 hr). Absolute levels of diabody uptake in CEA-positive xenografts were lower than minibody (4.37 ± 1.96 %ID/g at 4 hr; 2.47 ± 1.35 %ID/g at 18 hr) due to faster disappearance of the diabody from the circulation. Marked reduction of normal tissue activity (due to dehalogenation/metabolism of the radioiodinated proteins in liver, kidney, and other peripheral tissues) resulted in excellent tumor/normal ratios at the 18-hr time point, especially for the anti-CEA diabody. CEA positive tumors as small as 10 mg (<3 mm diameter) could be imaged, and I-124 anti-CEA minibodies demonstrated highly specific localization compared to FDG. Iodine-124 labeling of engineered antibody fragments provides a promising new class of tumor-specific probes for PET imaging of tumors and metastases.

Abstract ID: 450

Noninvasive Imaging of Protein–Protein Interactions in Living Animals

Gary D. Luker^1,2, Vijay Sharma^1,2 Christina M. Pica^1,2, Julie L. Dahlheimer^1,2, Wei Li^1,2, Joseph Ochesky^1,2, Christine E. Ryan^3,5, Helen Piwnica-Worms^3,4,5, David Piwnica-Worms^1,2

¹Molecular Imaging Center, Mallinckrodt Institute of Radiology, ²Department of Molecular Biology and Pharmacology, ³Department of Cell Biology and Physiology, ⁴Department of Internal Medicine, Washington University School of Medicine, St. Louis, MO 63110 USA, and ⁵Howard Hughes Medical Institute

Protein–protein interactions regulate critical pathways in cellular physiology and pathophysiology, including transcription, cell division and cell proliferation, signal transduction, oncogenic transformation, and cell death. Although protein interactions can be analyzed with a variety of methods in vitro and in cultured cells, interactions have not been investigated in intact, living animals. To enable noninvasive molecular imaging of protein–protein interactions in vivo, we engineered a fusion reporter gene comprising a mutant herpes simplex virus 1 thymidine kinase (HSV1-tk) and green fluorescent protein (GFP) for readout of a tetracycline-inducible two-hybrid system adapted for use with positron emission tomography (PET) and fluorescence microscopy. This novel imaging system was used to detect and quantify interactions between p53 tumor suppressor and an oncogenic binding partner, the large T antigen (TAg) of SV40 virus. To enable in vitro and in vivo two-hybrid assays, HeLa cells were transfected stably with the reporter gene under control of a Gal4 promoter. Hybrid proteins of Gal4-DNA binding domain-p53 (Gal4-DBD-p53) and VP16-TAg were used to drive transcription of the reporter gene in response to interacting proteins. As a negative control, coat protein (CP) from polyoma virus, which does not interact with p53, was fused to VP16. We used a bidirectional tetracycline-regulated promoter to allow inducible expression of pairs of hybrid proteins (Gal4-DBD-p53/VP16-TAg and Gal4-DBD-p53/VP16-CP) for the imaging two-hybrid system. In HeLa reporter cells stably transfected with a reverse tetracycline transactivator and Gal4-DBD-p53/VP16-TAg or Gal4-DBD-p53/VP16-CP, treatment with doxycycline for 48 hr in cell culture induced expression of hybrid proteins as determined by Western blotting. Interaction of Gal4-DBD-p53 and VP16-TAg in vitro increased activity of the HSV1-tk component of the reporter gene by approximately 76- and 190-fold as measured with substrates 8-³H-PCV and ¹⁸F-FHBG, respectively. Expression of the GFP component of the reporter gene was also detected by fluorescence microscopy. Conversely, induction of Gal4-DBD-p53 and VP16-CP did not significantly affect function of the reporter gene. To image protein interactions in vivo, we created tumor xenografts of HeLa cells transfected with either the interacting or control hybrid proteins. After treatment of mice with doxycycline, microPET imaging showed accumulation of ¹⁸F-FHBG only in tumors that expressed Gal4-DBD-p53 and VP16-TAg. Interaction of Gal4-DBD-p53 and VP16-TAg in vivo increased activity of the reporter gene by approximately six-fold above untreated or control Gal4-DBD-p53/VP16-CP tumors, as determined by region-of-interest analysis of microPET images and quantitative biodistribution studies with ¹⁸F-FHBG. Fluorescence microscopy of excised tumors revealed expression of GFP only in doxycycline-treated Gal4-DBD-p53/VP16-TAg tumors in vivo. These data demonstrate the feasibility of imaging protein–protein interactions in vivo and suggest that molecular imaging of protein interactions can be used for studies of functional proteomics in living animals.

Abstract ID: 452

Antivascular Effects of ZD6126 Assessed by Dynamic Contrast-Enhanced Magnetic Resonance Imaging (DCE-MRI)

Jeffrey Evelhoch¹, Patricia LoRusso¹, Zhanquan He¹, Zachary DelProposto¹, Kara Stark¹, Zahid Latif¹, Paula Morton¹, John Waterton², Catherine Wheeler², Alan Barge²

¹Karmanos Cancer Institute, Wayne State University, Detroit, MI 48201 and ²AstraZeneca, Alderly Park, UK

ZD6126 (N-acetylcolchinol-O-phosphate) is a novel vascular targeting agent which disrupts the tubulin cytoskeleton of neoendothelial cells, preferentially targeting tumor vasculature over established blood vessels. In order to optimally assess effects of ZD6126 in cancer patients, a method to serially evaluate changes in tumor blood supply is needed. We have used DCE-MRI to evaluate the antivascular effects of ZD6126 in mice bearing C38 colon adenocarcinoma. DCE-MRI studies were conducted prior to and 1 day after intravenous injection of ZD6126 at 200 mg/kg (n = 6), 100 mg/kg (n = 4), 50 mg/kg (n = 3), or 0 mg/kg (n = 4). High spatial resolution MRI data were acquired from the entire tumor volume every 6 sec before, during, and after intravenous bolus injection of Gd-DTPA. The signal increases were converted to contrast concentration and the initial (i.e., 60 sec) area under the curve (IAUC, which reflects both perfusion and permeability) was calculated on a pixel-by-pixel basis within regions-of-interest (ROI) including the entire tumor and muscle. The median muscle IAUC was used to account for differences in the arterial input function between studies. Both the 100-mg/kg [-82% ± 15% (SD)] and 200-mg/kg (−64% ± 20%) doses significantly decreased the median tumor IAUC, while 50 mg/kg produced an insignificant decrease (−24% ± 13%). Visual inspection of the images indicates that ZD6126 reduces the IAUC to negligible levels in the tumor center while a high IAUC rim remains after treatment. Although these preclinical results are promising, application of DCE-MRI in clinical trials of antivascular agents requires knowledge of its reproducibility to distinguish treatment effects from measurement errors. Thus, the reproducibility of the IAUC was evaluated in 19 patients (eight colorectal, three renal cell, two breast, two lung, one each of prostate, ovarian, sarcoma, and melanoma; tumor sites were: eight liver, five bone, three soft tissue, three pelvic) in two sequential studies separated by 2–5 days [mean 2.9 ± 1.1 days]. All patients were candidates for Phase I trials and had not received treatment during the 2 weeks prior to this study. Using methods similar to those used in the preclinical study, the IAUC was calculated on a pixel-by-pixel basis. There was no significant change between studies in the median tumor IAUC [paired t test, p = .47; mean change −3.8% ± 19.7%]. The overall test–retest coefficient of variation (CoV) was 13.9% for the median tumor IAUC in these Phase I patients, which compares favorably with the CoV for the control tumors in the preclinical study (25.0%). These results demonstrate that DCE-MRI, with reference to muscle, allows a reproducible assessment of tumor vascularity that should prove useful for clinical trials. Our initial results in the Phase I trial of ZD6126 demonstrate significant decreases in the tumor IAUC at doses as low as 56 mg/m². Overall, these results provide a clear demonstration that DCE-MRI provides a nonambiguous serial assessment of the effects of antivascular therapy on tumor vasculature in humans.

Abstract ID: 457

Imaging Firefly Luciferase Expression in the Lung in Live Animals After Intravenous Injection of Polyethylenimine/DNA Complexes

Jean-Luc Coll, Marie Favrot

INSERM EMI9924, Institut Albert Bonniot, La Tronche, 38706, France

Nonviral vectors based on polyethylenimine (PEI) are able to transfect a broad spectrum of cell lines in vitro, but transfect very preferentially lung cells in vivo after an intravenous injection of PEI/DNA complexes. This ability of PEI makes it an attractive vehicle for gene transfer into the lung, and more especially into Type II pneumo-nocytes. In order to determine the requirements for efficient PEI-mediated gene transfer, we use a luminescence imaging system, which allows the detection of cells and tissues that express firefly luciferase reporter gene in the living anesthetized animal. We present evidence that PEI-mediated gene delivery is indeed restricted to lung cells and that it is possible to quantify the level of gene expression.

Abstract ID: 459

¹⁹F NMR as a Tool to Measure the Efficiency of Adenoviral CD Gene Transfer by Monitoring 5-FC to 5-FU Conversion in Colon Adenocarcinoma

P. Kiss¹, A. Gustin⁴, D. Buchsbaum³, T. Simor¹, L. Lenard¹, S. Vickers⁴, G. A. Elgavish^1,2,5

¹Departments of Biochemistry and Molecular Genetics, ²Medicine, ³Radiation Oncology, ⁴Surgery and ⁵Center for Nuclear Imaging Research and Development, University of Alabama at Birmingham, Birmingham, AL

Introduction: Molecular chemotherapy is a novel approach to the treatment of gastrointestinal malignancies [1]. One strategy delivers an Escherichia coli gene that encodes the enzyme cytosine deaminase (CD) that catalyzes the conversion of an inactive prodrug, 5-fluorocytosine (5-FC), into the chemotherapeutic drug, 5-fluorouracil (5-FU). At UAB, the delivery is carried out via a replication incompetent adenoviral vector. 19F NMR spectroscopy has been used to monitor the conversion of 5-FC to 5-FU [2], but to date, no in vivo or in vitro study has been carried out to evaluate this method using an adenoviral CD vector. The aim of this study has been the use of 19F NMR to monitor the efficiency of adenoviral CD gene transfer as reflected in the conversion of 5-FC to 5-FU in a human colon adenocarcinoma cell line (LS174T) following infection at three different multiplicities of infection (MOIs).

Methods: Cells were infected with the adenovirus and placed either in plastic vials or injected into the flanks of nude mice. NMR measurements were carried out on a Bruker AM 360 spectrometer equipped with a surface coil probe. In vitro: 500 ml of 5-FC (0.05 mmol) was added to each vial and a spectrum collected every 5 min. In vivo: mice were anesthetized and placed in the probe with the tumor opposite the surface coil. After shimming, 500 ml of 5-FC was injected locally at the site of the tumor cell injection through a prepositioned silicone catheter. 19F spectra (64 scans each) were obtained.

Results: Results with MOI 150 showed a faster appearance of 5-FU and faster elimination of 5-FC than with MOI 75. Results with MOI 100 showed an intermediate rate both in vitro and in vivo. Thus, rate of conversion was MOI dependent most likely due to a dependence of CD protein concentration on MOI. These differences in protein concentration are related to the amount of CD gene delivered to the cancer cells. Increasing MOI increases the amount of gene delivery, thus increasing CD enzyme production.

Conclusions: This study demonstrates that NMR monitoring of CD-catalyzed drug conversion could be used as a measure of the efficiency of CD gene transfer. Correlation has been observed between the MOI and the rate of conversion of 5-FC to 5-FU both in vitro and in vivo. This correlation is simpler to interpret in the in vitro versus the in vivo experiments due to the additional compartments and kinetic pathways participating in emergence and clearance of both 5-FC and 5-FU in vivo versus in vitro.

Abstract ID: 462

Long-term Monitoring of Antigen-Specific In Vivo Labeled Cytotoxic T Lymphocytes (CTL)

Mikhail Doubrovin^1,*, Ekaterina Doubrovina^2,*, Julius Balatoni¹, Anna Ivanova¹, Vladimir Ponomarev¹, Richard O'Reilly², Ron Finn, Ronald Blasberg¹, Juri Gelovani¹

¹Department of Neurology and ²Department of Pediatrics, Memorial Sloan-Kettering Cancer Center, New York, NY 10021. ^*equal contribution

Rapid development of the therapeutic approaches based on the adoptive transfer of the tumor/virus-specific CTLs requires further progress in understanding the mechanisms and pathways involved in in vivo activation of the infused T cells and their migration to the specific targets. Passive in vitro labeling of CTLs with radioactive isotopes can be unstable and cannot be used effectively to study T-cell proliferation in the body. Extracorporal labeling of the CTLs stably expressing reporter gene has other disadvantages: (1) time-limited period for the imaging after labeling (half-life of the isotope), and (2) potential toxic effects related to the excessive radiolabeling. We present a novel method for noninvasive long-term in vivo imaging of the EBV-specific CTLs expressing hsv-1tk/egfp fusion transgene using the in vivo [124I]-FIAU and PET. EBV-specific CTLs were generated in vitro from the PBMC of the healthy donor stimulated with autologous EBV transformed B cell line (BLCL) and transfected with the hsv-1tk/egfp reporter gene using a GALV-based retrovirus. The transduced cells were positively selected for TK-GFP expression by FACS sorting (mean fluorescence of sorted population was 347). HSV-1 TK enzymatic activity was assessed in vitro using the [¹⁴C]FIAU-[³H]TdR radiotracer accumulation assay; K_i for FIAU in CTL-TG cells was 0.12 ml/min/g, which has been demonstrated to be sufficient for in vivo imaging. The sorted TK-GFP expressing CTLs retained the EBV-specific HLA restricted cytotoxicity as measured by Cr51 release assay, which was not affected by [¹²⁴I] and [¹³¹I] FIAU labeling in vitro. SCID mice were xenografted with four type tumors: autologous BLCL, allogeneic HLA matched BLCL, allogeneic EBV-negative B cell line, and HLA mismatched BLCL—in collateral sides of thighs and shoulders of SCID mice. When the tumors reached palpable size (~0.5 cm in diameter), TK-GFP-CTLs were administered intravenously at a dose 5 × 10⁷/mouse. [¹²⁴I]FIAU (~10 μCi/kg animal weight) was injected intravenously 24 hr after CTL administration. MicroPET monitoring started 2 hr after FIAU injection. Specific accumulation of radiotracer was observed only in autologous and HLA-matched EBV tumors and in the spleen. Sequential imaging showed a gradual increase in tumor FIAU signal, demonstrating progressive targeting of TK-GFP-CTLs after intravenous administration over the 72-hr period. The FIAU signal decreased after 7 days due to the FIAU clearance. No accumulation was observed in the control tumors. The microPET observations were confirmed by tissue sampling; radioactivity levels at 24 hr were 0.23%, 0.19%, 0.03%, and 0.01% dose/g in tumors (1) through (4), respectively. Fluorescence microscopy demonstrated presence of eGFP expressing cells in tumors (1) and (2), and in the spleen, but not in tumors (3) and (4). This allows for in vivo labeling and noninvasive monitoring of the reporter gene transduced CTLs distribution and trafficking in the living organism over the long period of time is feasible with [¹²⁴I]-FIAU and PET.

Abstract ID: 464

Endometase/Matrilysin-2 in Human Endometrial, Breast, and Prostate Cancers

Hyun I. Park, Yunge Zhao, Tiebang Kang, Elena Filenova, Fuli Yu, Martin A. Schwartz, Qing-Xiang Amy Sang

Endometase, matrix metalloproteinase 26 (MMP-26/matrilysin-2) was cloned from human endometrial tumor cDNA library. It may be a novel marker for carcinoma diagnosis and a new target for cancer treatment. Its mRNA was specifically expressed in uterus and many types of carcinomas (cancers of epithelia), however, it was not expressed in most normal human tissues/organs tested. Its zymogen with a molecular mass of 28 kDa was expressed and isolated from Escherichia coli. The folded endometase from inclusion bodies could be activated by dialysis. Endometase digested Type-I gelatin, alpha-1-proteinase inhibitor, vitronectin, fibronectin, and fibrinogen. It also cleaved many peptide substrates including MMP substrates and a substrate for a tumor necrosis factor-alpha converting enzyme. It may play a role in angiogenesis and blood coagulation system. The pH study revealed two pK_a values (4.8 and 9.5). The enzyme was inhibited by tissue inhibitor of metalloproteinases-1, −2, and −4 and our new synthetic MMP inhibitors with K_i values at pM–nM range. Inhibitor mechanistic and kinetic studies using our mechanism-based inhibitors revealed that endometase has a big deep S1′ active site pocket. Four specific antibodies were developed. The endometase protein was expressed in human endometrial, breast, ovarian carcinoma, and advanced androgen-repressed prostate carcinoma cells. Endometase may play important roles in human reproductive system, in blood coagulation and angiogenesis, and in cancer progression. [Supported by NIH, NSF, and American Cancer Society.]

Abstract ID: 465

Scanning Electrochemical Microscopy of Metastatic Human Breast Cells

Susan A. Rotenberg, Biao Liu, Michael V. Mirkin

Department of Chemistry and Biochemistry, Queens College, City University of New York, Flushing, NY

Recent developments from this laboratory indicate that metastatic human breast cells exhibit lower redox activity than normal human breast cells. This work applies the scanning electrochemical microscope (SECM) to cultured monolayer breast cells. Redox activity is measured in a single cell by monitoring electrochemical changes with an ultramicroelectrode, which records redox changes made by the cell to a membrane-diffusible quinone chemical mediator (e.g., menadione or 1,2-napthoquinone). The electrode is positioned closely to the outer surface of the plasma membrane without penetrating it. In freely traversing the plasma membrane, the chemical mediator is first converted to the diol form by the electrode and enters the cell in this state; while in the cell, the mediator undergoes conversion to the dione by an undefined intracellular oxidant, and upon exiting the cell, is restored to the diol by the electrode thus resulting in a signal. The chemical mediator thus serves as an effective reporting device for intracellular redox reactivity occurring within a specific range of reduction potentials. An important application of the SECM is that it can be used to scan the redox reactivity of the entire surface of a single cell, thus yielding a redox map of the cell. Redox reactivity has been measured in nontransformed, nonmotile human breast MCF-10A cells, and compared with the same cells in which bovine protein kinase Cα (PKCα) was genetically engineered, and with authentically metastatic human breast cells (MDA-MB-231 cells) that express high endogenous levels of PKCα. High-level expression of PKCα produces an array of phenotypes that include exaggerated migration, decreased proliferation (Cell Growth and Diff 10: 343–352, 1999), and decreased redox reactivity as measured by SECM (Proc Natl Acad Sci 97: 9855–9860, 2000). The SECM has been used to detect metastatic cells in a mixed monolayer field of nonmetastatic breast cells, and potentially can be extended to detection of micrometastases in breast tissue biopsies. The structural requirements of PKCα that enable its effect on the SECM signal were explored by comparing the ability of site-directed mutants to produce these phenotypes in MCF-10A cells. Those mutants tested included: (1) a kinase-defective dominant-negative (DN) mutant (K368R), (2) a catalytically competent mutant in which the RACK-binding site (aa 218–226) was deleted, and (3) a catalytically competent mutant in which the PICK1-binding site (aa 669–672) was deleted. The RACK and PICK sites are important determinants of PKCα subcellular localization. Following transient transfection of each cDNA construct, expression of wild-type and mutant PKCα proteins could be demonstrated by Western blot. Phenotypic characterization of transfectant cells confirmed that wild-type PKCα decreases redox reactivity, whereas the three PKCα mutants were ineffective and equivalent to the vector control. It is concluded that redox reactivity is mediated by PKCα that retains catalytic function and is capable of localizing to specific subcellular sites. [Funded by NIH CA91341.]

Abstract ID: 468

FDG-PET Scanning Evaluation in Phase I Study of Flavopiridol

Kenji Kawada, Hironobu Minami, Kohji Murakami, Noriko Usubuchi, Michinori Ebi, Shin Tahara, Kenichi Ishizawa, Tadahiko Igarashi, Kuniaki Itoh, Yasutsuna Sasaki

National Cancer Center Hospital East, Kashiwa, Japan

Background: FDG-PET scanning is a metabolic imaging procedure that has become a useful technique for monitoring chemotherapy response in cancer. Flavopiridol is a cyclin-dependent kinases inhibitor that has undergone Phase I study in Japan.

Purpose: To assess the effects of flavopiridol by means of FDG-PET scanning.

Methods: Flavopiridol was administered as a 24-hr intravenous infusion on Days 1, 8, 15, and 22, with cycles repeated every 5 weeks. CT and FDG-PET scanning were done before and after the first cycle of flavopiridol administration. For quantitative evaluation of FDG-uptake in tissue, a small square tissue sample (4 by 4 pixels, approximately 1.6 by 1.6 cm) was placed in the high focal uptake value areas on the scanner. Standard uptake values (SUV) were measured within this region. Measurements of the greatest major axis of tumors were done by CT scanning.

Result: Five patients (two colon cancer, two NSCLC, one rectal cancer) were evaluated for responses by both CT and FDG-PET scanning. FDG-PET scanning has the tendency to detect a decrease of metabolic activity in tumor tissue even though no tumor responses can be detected by CT scanning.

Conclusion: Results suggest that FDG-PET scanning might be useful in evaluating tumor response to flavopiridol. Observation of a homogeneous population over a long period is essential to assess the effects of cytostatic agents such as flavopiridol by means of FDG-PET scanning. [Supported by Aventis Pharma.]

Abstract ID: 469

Imaging of Luciferase Transfected Human Tumors in Nude Mice

Gisela Caceres¹, XiaoYun Zhu², Ralitza Zamkina¹, Mike Dauphinee¹, Peter Andreotti³

¹Rumbaugh-Goodwin Institute, ²Sunol Molecular Corporation, and ³ASD Laboratories, Florida, USA

Bioluminescent imaging has been used to study the growth and metastasis of luciferase-transfected human tumors in nude mice. Growth and metastasis of the MCF-7 human breast cell line transfected with the Photinus pyralis luciferase gene were qualitatively and quantitatively analyzed in nude mice using the Night Owl Molecular Light Imager. Studies with 10⁶–10⁷ cells per mouse showed differences in tumor distribution in vivo depending on the route of inoculation. Cells injected intravenously were not detectable until metastasis was observed in lymph nodes after 5–6 weeks. Cells injected intraperitoneally were localized in the peritoneal cavity with spleen infiltration. Cells injected subcutaneously could be detected by imaging prior to observing a measurable tumor mass. Similar studies have been performed with transfected DU-145 prostate cells.

Abstract ID: 470

Novel Far Red and Near Infrared Dyes for Cell Tracking In Vivo

Brian Gray¹, N. Askenasy², E. Breslin¹, K. Muirhead³, B. Ohlsson-Wilhelm³, D. L. Farkas²

¹PTI Research, Inc., Exton, PA, ²Carnegie Mellon University, Pittsburgh, PA, and ³SciGro, Inc., Malvern, PA

The activity of critical genes/gene products involved in cellular therapies may be exquisitely sensitive to signals in the local environment at different sites within the body. Thus, it is useful to have separate extrinsic markers that can be used to follow the trafficking of such cells within the body, in order to determine if/how site of localization affects expression levels for genes of interest. Constitutively expressed genetic markers, such as GFP, are not easily introduced into all cells and may not readily allow assessment of cell function and/or proliferation. Stable visible fluorescing membrane dyes (e.g., PKH2, PKH26, or PKH67) may be used to label any cell and are apportioned equally to daughter cells upon division. They have been proven useful for monitoring which subpopulations of cells proliferate in response to a given stimulus and extent of cell division. We report here on the development of a new series of fluorescent membrane dyes, emitting in the far red (e.g., PTIR271 and PTIR272) and near infrared (e.g., PTIR273 and PTIR274). All dyes are suitably bright at concentrations nontoxic for cellular labeling, and are well retained by cells. Thus, these dyes are good candidates as markers to locate and track therapeutic cells and monitor their activities. The far red emitters (665–720 nm) are suitable for monitoring cells in blood or bone marrow using four-color flow cytometry. Both PTIR271 and PTIR272 are spectrally compatible (nonoverlapping) with fluorescein and phycoerythrin, thus requiring little or no compensation [1] for quantitation. PTIR273 is also detectable but suboptimally excited. The near-infrared (814–840 nm) emitting PTIR membrane dyes are potentially excellent for in vivo imaging, and spectrally complementary to/synergistic with other tags used for this purpose (constitutive genetic markers [2], [3] and fluorescent antibodies [4]). In preliminary studies using a custom imaging system developed at CMU [4], good signal-to-noise transdermal fluorescence images were obtained from labeled YAC-1 cells implanted subcutaneously in nude mice. Results to date suggest that these new dyes: (1) may enable detection of tagged cells and associated gene or protein markers at previously unattainable depths in tissues; (2) will extend the range of probes available for multicolor cellular analysis both in vitro and in vivo; and (3) can be highly useful as tools for noninvasive assessment of mechanisms involved in cell-based therapy.

Abstract ID: 471

The Generation of a Conditional Luciferase Reporter Mouse Intended to Enable Bioluminescence Imaging of Spontaneous Tumorigenesis

Scott Lyons, Paul Krimpenfort, Ralph Meuwissen, Anton Berns

Department of Molecular Genetics, The Netherlands Cancer Institute, Plesmanlaan 121, 1066CX, Amsterdam, The Netherlands

As part of an effort to render tumors developing from conditional (Cre/LoxP mediated) mouse tumor models suitable for bioluminescence imaging, we report the generation of a new transgenic reporter mouse that can express luciferase in all tissues only after Cre-mediated recombination. Spontaneous mouse cancer models are well suited for stringent preclinical evaluation of anticancer therapies. However, the stochastic nature of tumor onset and progression in these models, especially when occurring at deep-tissue sites, necessitates assumptions regarding tumor burden before and after treatment and consequently, the participation of large animal cohorts to generate meaningful data. Non-invasive imaging of spontaneous tumorigenesis would largely obviate such limitations, as the timing of drug administration could be accurately synchronized with tumor mass without assumption, and tumor response to intervention could also be directly visualized in vivo. The conditional luciferase reporter construct comprises a Beta-actin promoter (for ubiquitous transgene expression) upstream of a “floxed” green fluorescent protein (GFP) cassette and a promoterless luciferase cassette positioned immediately downstream of GFP. Thus, in the absence of Cre recombinase activity, cells only express GFP and not luciferase. Cre-mediated deletion of the “floxed” GFP cassette, however, places the luciferase cassette adjacent to the Beta-actin promoter and results in luciferase expression. We show with this reporter that in the absence of Cre recombinase activity, all organs analyzed emit very low to negligible amounts of light (as measured by IVIS, Xenogen, Alameda). After crossing to a ubiquitously expressing Cre transgenic line however, all organs became 5 to 6 orders of magnitude brighter per milligram of wet tissue weight. We also demonstrate that luciferase expression in this conditional reporter line can be switched on somatically following adenoviral Cre infection or by crossing with a tamoxifen-inducible Cre line. The amount of light emitted from switched tissues is comparable to that of an existing luminescent spontaneous tumor mouse model, used previously by us for tumor intervention studies (manuscript submitted). Therefore, as the overall level of measured background luminescence from unswitched tissue is low, we propose that this luciferase reporter line will enable bioluminescence imaging of spontaneous tumorigenesis when combined with other conditional (Cre/loxP dependent) mouse cancer models.

Abstract ID: 472

Whole Body Optical Imaging and Therapeutic Rationale for Experimental Bone Metastases Clemens

Lowik¹, Gabri van der Pluijm¹, Marielle Kroon¹, Ivo Que¹, Jeroen Buijs¹, Antoinette Wetterwald², Ekaterina Gresko², Marco Bisoffi², George Thalmann², Marco Cecchini²

¹Department of Endocrinology, Leiden University Medical Centre, Leiden, The Netherlands, and ²Gene Therapy Laboratory, Department of Clinical Research and Department of Urology, University of Bern, Bern, Switzerland

The development of novel strategies for cancer treatment requires the elaboration of sensitive and noninvasive in vivo assays to monitor their effectiveness. Recently, we developed an animal model, based on injection of luciferase-transfected cancer cells into the left cardiac ventricle of mice and whole body bioluminescent reporter imaging (BRI), which closely mimics micrometastatic spread. BRI can detect microscopic bone marrow metastases of about 0.5 mm³ volume. This sensitivity translates into early detection of intramedullary tumor growth, preceding the appearance of a radiologically evident osteolysis by 2 weeks. This model was used for investigating the effectiveness of therapeutic strategies aiming to repress tumor-induced osteolysis and/or angiogenesis. The bone metastatic clone MDA-231-B, derived from the MDA-MB-231 mammary carcinoma cell line and stably transfected with the pCMV plasmid containing the firefly luciferase gene (MDA-231-B/Luc⁺), were either inoculated into the left cardiac ventricle (ic) or implanted subcutaneously (sc) in BALB/c nu/nu mice. Development of bone metastases and subcutaneous tumor growth was monitored in vivo by BRI. Extent of tumor-induced osteolysis was analyzed by conventional radiography. Distinct photon emission localized at the metaphyses of the hind legs and spine was first detected 2 weeks after the intracardiac injection of MDA-231-B/Luc⁺ cells. Localization at these specific sites is consistent with the hypothesis that cancer cells colonize electively sites of active marrow hemopoiesis and bone remodeling. At later time points, metastases were also detected in the calcaneus bones. BRI revealed that in transgenic mice expressing luciferase under the control of the osteocalcin promoter, a marker of osteoblast activity, the calcaneus bone is also a site of active bone remodeling. The bisphosphonate olpadronate, a strong inhibitor of bone resorption, given continuously from first appearance of bone metastases, did not inhibit tumor growth, as detected by BRI, while strongly inhibited tumor osteolysis, as detected by radiography. In contrast, continuous treatment with endostatin, a known inhibitor of angiogenesis, arrested further growth and induced, within a week, regression of subcutaneous tumors, as detected by BRI and tumor size. BRI is an extremely useful method for evaluating the efficacy of therapeutic strategies aiming to repress bone metastatic development and/or growth. Association of antiangiogenic and antiosteolytic regimens seems to be a good rationale for the treatment of bone metastases.

Abstract ID: 474

A New Technique to Identify Putative Diagnostic or Therapeutic Molecular Imaging Targets: Ligand-Enhanced MALDI-MS Proteomics

S. K. Hobbs, G. Shi, M. D. Bednarski

Lucas MRS Center, Department of Radiology, Stanford Medical Center, Palo Alto, CA 94304

High-throughput, small-molecule screening is currently being developed for new drug discovery and for investigating the ability to bind proteins of interest or the ability to perturb a specific biological pathway. The design of receptor-based molecular imaging agents is based upon the attachment of these small molecules or ligands to a linker and then to an imaging probe. We describe here a new strategy to synthesize libraries of ligand–linker conjugates and to display them at an interface for proteomic analysis. This new technique requires no knowledge of a particular protein target but instead finds proteins in a tissue sample that bind to known small molecular ligand–linker conjugates that can then be used for designing molecular imaging agents for specific pathologies. Molecular specificity and sensitivity of matrix-assisted laser desorption/ionization mass spectroscopy (MALDI-MS) lends itself to analyze proteins and peptides on surfaces. We modified the surface of the commercially available MALDI-MS plates with thin-film hydrogel monolayers containing ligand–linker conjugates. The advantage of these surfaces is their easy reproducibility, modifications, and low nonspecific binding of nontarget proteins to the monolayer surfaces. The ligand–linker conjugates can easily be introduced into the thin films at varying concentrations. We explored the feasibility of this film technology of a ligand-conjugate hydrogel MALDI-MS surface by investigating protein–ligand interactions using a variety of ligands and crude cell lysates. We have shown: (1) concentration-dependent protein binding to immobilized ligands, (2) competitive binding inhibition, and (3) specific concentration-dependant protein binding to immobilized ligand within a complex cell lysate. Using this technique, we have utilized a peptidomimetic ligand–linker conjugate designed to bind cell adhesion molecules to image murine tumor models with MRI. Linker–ligand conjugate arrays of immobilized small molecules within organic thin-film surfaces are a new system for screening for molecular imaging targets. It is unnecessary in this system to have a predetermined protein target, but instead, it takes advantage of small molecule libraries as a means to discern new biological markers of disease.

Abstract ID: 475

Dynamic Magnetic Resonance Imaging of Prostate Cancer in Mice

Renshu Zhang¹, Ercheng Li¹, Yusuf D. Ali¹, Huafu Song¹, Kuang J. Fan², Kathleen F. Pirollo³, Esther H. Chang³, Paul C. Wang¹

Department of Radiology and Cancer Center, Department of Pathology, Howard University, Washington, DC, 20060, and Department of Oncology, Georgetown University Medical Center, Washington, DC 20007

Prostate cancer is the most common malignant tumor and is the second most common cause of cancer-related mortality among men. Prostate cancer patients often require frequent imaging follow-up of their tumors. The noninvasive and lack of radiation involvement with Magnetic Resonance Imaging (MRI) make it an ideal choice of imaging method for routine tumor monitoring. In this study, we performed a dynamic contrast-enhanced MRI study of prostate tumors in athymic nude mice to characterize heterogeneity of the tumor, to assess the kinetics of contrast enhancement, and to correlate the magnetic resonance (MR) contrast enhancement with histopathological findings. 10⁷ Prostate cancer cells (DU145) suspended in Matrigel were injected subcutaneously into the lower back near the tail of the mice. The MRI scan was performed when the tumors grew to approximately 1.0 cm. A catheter was placed in the jugular vein enabling administration of the contrast agent without changing the animal's position within the magnet. During MR scanning, the mouse was anesthetized with isoflurane through a nasobuccal cone mask. The mouse was put in the lateral decubitus position, inside the magnet (a Varian 4.7 T horizontal bore NMR machine). A 2.0-cm single-loop RF coil was placed around the tumor for imaging. T1-weighted spin-echo imaging technique with echo time 12 msec and repetition time 750 msec was used. The image slice thickness was 1 mm, field-of-view 2.0 cm, and matrix size 256 × 200. A set of coronal and transverse images was taken as baseline before injection of the contrast agent, Magnevist (Berlex Laboratories) at 0.4 mmol/kg. A series of fast spin-echo images were acquired during the injection of the contrast agent. A series of T1-weight spin-echo images were obtained every 20 min after the completion of injection to monitor the dynamic changes of contrast enhancement. The early images from the first 20 min showed a rapid contrast enhancement in the periphery of the tumors. The later images showed progressive uneven enhancement throughout the tumor. After the MR scanning, tumors were dissected to match the image slice. In the histological specimens, the highly vascular areas correspond well to the contrast-enhanced areas in the images. The nonenhanced areas indicate necrotic and cystic lesions. The rate and the overall steady state of contrast enhancement have been used to determine the malignancy of tumors. The dynamic contrast-enhanced MR images can be used as a powerful means to characterize tumor prognosis, and to monitor therapeutic responses to chemo, hormonal, and radiotherapy.

Abstract ID: 477

Direct Quantitative Imaging of Trace Metal Distribution with 200-nm Resolution

Barry Lai¹, Zhonghou Cai¹, Jorg Maser¹, Dan Legnini¹, Peter Ilinski¹, Stefan Vogt¹, Anton P. J. Stampfl^1,2

¹Advanced Photon Source, Argonne National Laboratory, Argonne, IL 60439 and ²Physics Division, Australian

Nuclear Science and Technology Organisation, Lucas Heights, NSW, 2234, Australia

Metal compounds can induce diverse biological effects, for example, Cr(VI) complexes are carcinogens while cisplatin is a common chemotherapeutic drug. Thus, the ability to directly image metals (usually presented in trace quantity) and their subcellular distribution is important. A new noninvasive microscopy technique based on a scanning X-ray fluorescence microprobe has been developed with the following characteristics: (1) it enables any element above silicon, including most metals, to be directly imaged without any staining or enhancing agent. In fact, this technique had been used for studying natural bacteria in a hydrated state. (2) For most transition and heavy metals, trace concentrations of < 1 ppm are detectable, while more than 10 different elements can be mapped simultaneously. (3) The demonstrated spatial resolution is 200 nm. (4) It can also reveal the chemical states, such as Cr(VI) versus Cr(III), Pt(IV) versus Pt(II). Initial applications of this technique included determining the intracellular biotransformation products of chromium carcinogens and metal-containing drugs on individual cells. The high elemental sensitivity of this technique allowed these studies to be performed at naturally occurring concentrations or at clinically relevant doses without the need to artificially elevate the concentrations. [This work supported by the U.S. Department of Energy under contract no. W-31-109-Eng-38.]

Abstract ID: 478

Magnetic Resonance Imaging and Gene Expression Changes in Focused Ultrasound-treated Tumor Model

W. Hundt¹, E. L. Yuh¹, N. T. Sanghvi², R. Seip², M. D. Bednarski¹

¹Department of Radiology, Stanford University, Stanford, CA, and ²Focus Surgery, Indianapolis, IN

Introduction: Imaging methods are being developed to follow thermal therapies for tumor ablation in cancer. MRI can be used to monitor temperature [1], but gives little information about molecular changes. The search for molecular targets and development of molecular imaging agents is necessary to elucidate molecular changes in order to monitor thermal therapies. Focused ultrasound (FUS) treatment of a mouse squamous cell carcinoma (SCC) provides a model system to study gene expression patterns at different levels of energy deposition [2]. Like rf ablation, radiotherapy, and cryotherapy, FUS causes tissue changes through energy deposition. We performed functional genomics on tumor tissue treated with various energy levels using FUS to define targets for the development of molecular imaging agents.

Materials and Methods: FUS was applied to 1 cm intradermal SCC VII tumors in C3H/Km mice using a dual ultrasound imaging (6 MHz)/therapeutic (1 MHz) system. Three dose regimes were investigated. The low and intermediate doses consisted of 50 msec FUS pulses with intensity 2000 W/cm² and pulse frequency of 0.5 Hz, applied for 5 and 10 min, respectively. The high dose consisted of continuous-wave FUS applied for 20 sec. Pre- and postcontrast T1- and T2-weighted images of FUS-treated tumors and untreated control tumors on the opposite flank were obtained on a Varian 4.7-T MRI scanner. After FUS treatment and imaging were completed, total RNA was isolated from tumors. Labeled cRNA probes were synthesized from the total RNA, and hybridized to a mouse genome oligonucleotide microarray (Affymetrix; Santa Clara, CA). Each gene was analyzed for fold change (FC), or factor by which a gene's mRNA product was elevated in treated tumors over their untreated controls. We analyzed eight pairs of FUS-treated tumors and untreated controls. Genes up-regulated more than two-fold in all tumors receiving similar FUS doses were further investigated.

Results: T1-weighted pre- and postcontrast and T2-weighted MRI images showed homogeneous intensity throughout all FUS-treated tumors and their untreated controls. No evidence of tissue changes after FUS treatment was seen in any of the MRI images. Despite a lack of MRI changes, gene expression analysis revealed profound changes in expression levels of four genes in the high-dose FUS-treated tumors. The highest up-regulation occurred for HSP 70 and a related MHC Class III gene (FC = 14.0 and 27.7, respectively). Other consistently up-regulated genes encoded HSP40 and an insulin-like growth-factor binding protein. Similar changes were observed in tumors treated with lower FUS doses.

Conclusion: Using FUS treatment of SCC tumors as a model system, we demonstrated dramatic changes in gene expression upon initiation of thermal therapy. These early changes are not visualized using standard MRI sequences. Functional genomic analyses of treated tumors revealed four potential targets for molecular imaging probes. These targets seem to be up-regulated prior to any visible MRI changes. Development of MRI probes utilizing these targets is in progress.

Abstract ID: 481

In Vivo Model of Tumor-Targeted Gene Expression Using Small Animal Bioluminescent Imaging

F. Diehn¹, N. G. Costouros², M. S. Miller², K. C. P. Li³, H. R. Alexander², S. K. Libutti²

¹HHMI-NIH Research Scholars Program and Diagnostic Radiology, Clinical Center, ²Surgery Branch, National Cancer Institute, and ³Imaging Sciences and Diagnostic Radiology, Clinical Center, National Institutes of Health, Bethesda, MD

Introduction: Real-time bioluminescent imaging has emerged as a powerful tool for assaying molecular and cellular processes as they occur in vivo. Our purpose was to take advantage of this technology in order to establish a mouse model system for visualizing the targeting of antitumor gene therapy to neoplastic cells.

Methods: C57/BL6 mice were injected with 1 × 10⁶ murine colon adenocarcinoma cells (MC38) via subcutaneous or intraperitoneal routes. After 9 days, 1 × 10⁸ pfu of a recombinant, thymidine kinase deleted vaccinia virus expressing luciferase was administered intraperitoneally. Four days after virus injection, d-luciferin (150 μg/g body weight) was administered intraperitoneally to ketamine/xylazine-anesthetized animals. The mice were placed in a dark box and a reference grayscale image was obtained under low-light conditions. Five minutes after luciferin injection, bioluminescent imaging was performed with integration times ranging from 1 to 20 min. All images were obtained using an ORCA-2 digital CCD camera (Hamamatsu). Subsequent image analysis was performed on a Macintosh using Openlab software (Improvision).

Results: Using small animal whole-body bioluminescence, we found that vaccinia virus encoding luciferase tracked specifically to tumors. In mice receiving subcutaneous tumor cell injections, high levels of bioluminescence were observed at the site of the tumors, while the peritoneum and abdominal organs did not harbor detectable luciferase activity. Conversely, mice receiving intraperitoneal tumor cell injections did display strong luciferase activity in the peritoneal cavity.

Conclusions: This animal model validates the ability of our in vivo small-animal, whole-body bioluminescence station to noninvasively assay the specific targeting of genetic vectors to tumor cells. Other investigators have previously employed bioluminescent imaging to demonstrate in utero gene therapy, quantify antineoplastic drug therapy, perform transgene analysis, or model localized gene therapy in normal tissue. Thymidine kinase deleted vaccinia virus has been shown previously to have a strong tropism for tumors. Using this recombinant vaccinia virus expressing the reporter gene luciferase, we show that real-time bioluminescent imaging can demonstrate the targeted delivery of genetic material to neoplasms. This viral delivery system serves as a model for a variety of gene therapy delivery vectors that could potentially target neoplastic cells. We plan to evaluate both the ability of various gene therapy delivery vectors to localize to tumors and the ability of various therapeutic genes to hinder tumor growth. Further work will also include extending this approach to lung and liver metastasis models.

Abstract ID: 483

Improved Cancer Immunoscintigraphy with Bispecific Antibody (BsMAb) Pretargeting Systems

R. M. Sharkey¹, W. J. McBride², H. Karacay¹, E. Rossi³, K. Chang³, G. L. Griffiths², H. J. Hansen², D. M. Goldenberg¹

¹Garden State Cancer Center, Belleville, NJ, ²Immunomedics, Inc., Morris Plains, NJ, and ³IBC Pharmaceuticals, LLC, Morris Plains, NJ

Several BsMAb pretargeting systems have been developed that can be used in a variety of imaging applications including external scintigraphy and even PET. BsMAb have been prepared by chemical conjugation or by recombinant molecular engineering of an anti-carcinoembryonic antigen (CEA) antibody (hMN-14) paired with antibodies directed against indium-loaded DTPA (734) or histamine–succinyl–glycine (HSG; 679). Pretargeting studies in tumor-bearing mice using the hMN-14 × 734 F(ab′)₂ BsMAb with a divalent DTPA peptide containing a ^99mTc-binding chelate achieved tumor/blood and tumor/kidney ratios of 7:1 and 5:1, respectively, within 3 hr of the peptide's injection, compared to a directly ^99mTc-hMN-14 Fab′ with no tumor/organ ratio exceeding 1:1 at the same time. A recombinant-engineered BsMAb composed of scFv from the anti-CEA and anti-HSG antibodies has shown superior tumor/nontumor ratios compared to a chemically conjugated anti-CEA × anti-HSG F(ab′)₂ using an ¹¹¹Inlabeled divalent HSG peptide, suggesting that engineered BsMAbs may provide additional advantages for pretargeting imaging applications. The HSG pretargeting system represents a significant advancement because of its flexibility; unlike other BsMAb systems, the anti-HSG antibody is not responsible for binding the specific effector molecule. Thus, divalent HSG-containing peptides can be synthesized to bear a variety of ligands to optimize radionuclide binding; for example, HSG peptides capable of binding ^99mTc and ¹¹¹In have been prepared with similar in vivo targeting properties as the other pretargeting system. With superior targeting ratios as compared to directly radiolabeled antibodies, the use of engineered humanized antibodies to eliminate agent immunogenicity, and an ability to adapt this system to use in immunoscintigraphy with SPECT or PET agents, these BsMAb pretargeting systems may provide next-generation imaging modalities for improved cancer detection and diagnosis. [Supported in part by PHS grant CA84379 and DOE grant FG01-00NE22941.]

Abstract ID: 484

Novel Transporter for Gadolinium Enables Fast and Specific Cellular Uptake Into Tumor Cells

S. Heckl, J. Debus, J. Jenne, R. Pipkorn, R. Rastert, W. Waldeck, G van Kaick, K. Braun

Deutsches Krebsforschungszentrum Heidelberg

Introduction: At present, the most commonly used contrast agents are Gadolinium-based compounds that accumulate in the interstitial space and are unable to pass the cellular membrane. In this study, a transmembrane transport system of human origin was coupled to the Gadolinium complex to enable passage into the intracellular space and to differentiate between tumor and nontumor cells.

Materials and Methods: Melanoma, C6-glioma, and HeLa-Cervix-carcinoma cells as well as nontumor cells (lymphocytes) were incubated with a Gadolinium complex, which was, in turn, linked to a target-cell-specific peptide module (100 pM). In order to improve the efficiency of cytoplasmatic transport, a transmembrane transport system (human origin) was cleavably covalently bonded to the Gd complex peptide module. A Gd complex without transporter (Magnevist) was used as a control. Influx and efflux were measured every 10 min with MRI (1.5 T; Siemens Magnetom Vision Plus; circular polarised head-coil). Additional measurements were performed using confocal laser scanning-microscopy.

Results: After just 10 min, Gd complex peptide modules could be detected in tumor and nontumor cells. Gd complexes without transporter did not enter the intracellular space. The Gd complex peptide-module was transported rapidly out of nontumor cells but remained in tumor cells due to target specificity. Apoptosis was not detectable.

Discussion: A rapid and specific uptake of gadolinium into the cytoplasm of tumor and nontumor cells is possible. The complex peptide module persists only in the specified tumor cells and thus allows a demarcation of tumor versus nontumor cells. Intraoperative MRI in neurosurgical procedures does not allow a clear delineation of malignant versus healthy tissue due to the outflow of the interstitial contrast agent Magnevist from the opened interstitium. This Gd complex peptide module could be an elegant solution.

Abstract ID: 485

Genetic Profiling of Heterogenous Tumor Regions Defined by Dynamic Contrast-Enhanced MRI for the Identification of Molecular Targets

N. G. Costouros¹, M. S. Miller¹, D. Lorang¹, Y. Zhang², F. Diehn², H. R. Alexander¹, P. L. Choyke², M. V. Knopp³, S. K. Libutti¹

¹Surgery Branch, National Cancer Institute, National Institutes of Health, Bethesda, MD, ²Department of Radiology, Clinical Center, National Institutes of Health, Bethesda, MD, and ³Department of Radiology, Ohio State University

We have developed a murine tumor model with the capability for genetic profiling of heterogenous tumor regions distinguished by pharmacokinetic analysis based on dynamic contrast-enhanced MRI (DEMRI). In contrast to normal vasculature, tumor vessels tend to form disorganized networks and contain an incomplete endothelial layer. Based on these structural abnormalities, differences in tumor endothelium can be exploited using DEMRI. Such analysis is useful for the investigation of tumor angiogenesis. Furthermore, with increased interest in anti-angiogenic tumor therapies, and the first clinical trials using such agents in progress, it is imperative that new methods be developed to measure clinical responses. C57BL/6 mice are inoculated in the right flank with 10⁶ MC38 murine colon adenocarcinoma cells and tumors are allowed to grow to approximately 1 cm in diameter. Mice are imaged axially using a 4.7-T Bruker (Billerica, MA) Avance MR scanner using a T1-weighted gradient-echo sequence. Gd-DTPA contrast is administered through a cannulated tail vein over 1 min using a mechanical pump injector and images are acquired every 20 sec for 20 min. Pharmacokinetic analysis is performed using software written in IDL by MVK based on the two-compartment pharmacokinetic model proposed by G. Brix (1991). Time–activity curves are generated for different regions-of-interest (ROI) and fitted to the pharmacokinetic model. Tumors are harvested, sectioned axially, and coregistered with the corresponding DEMRI slices. PCNA, caspase-3, and CD31 immunohistochemistry is performed on histologic sections. Tumor ROIs that differ by pharmacokinetic analysis are selectively dissected using laser capture microdissection and RNA is extracted. mRNA was amplified using a modified Eberwine technique and ROIs were compared using cDNA glass slide microarrays. On gross histology and CD31 staining, areas of rapid enhancement display larger, more mature vessels than slow enhancing regions. PCNA and caspase-3 staining show increased proliferation and apoptosis in the periphery, possibly indicating active tumor proliferation and selective turnover in less dense peripheral regions that receive a good vascular supply, and sessile tumor cells centrally that have increased contact inhibition and poor vascular supply. Microarray analysis shows differences in gene expression between rapidly and slowly enhancing ROIs. Further analysis is needed to conclude significant expression patterns. Coupled with immunohistochemistry and DEMRI findings, gene expression profiling may help elucidate mechanisms of treatment responses after administration of specific therapies and may provide a basis to deduce molecular activity from DEMRI findings. Furthermore, genetic profiling of different tumor regions may provide insight into new targets for therapy and novel contrast agent development.

Abstract ID: 487

In Vivo Monitoring of PC 3 Induced Prostate Tumor Response to Taxotere in Nude Mouse Using Intracellular Sodium Magnetic Resonance Imaging, 18-FDG Positron Emission Tomography, and Immunohistological Characterization

S. R. Sharma¹, P. Esser¹, H. Van Heertum¹, P. J. Cannon², J. K. Katz¹

Departments of ¹Radiology and ²Medicine, College of Physicians and Surgeons, Columbia University, New York, NY 10032

Tumor characterization by in vivo (Na)-MRI and 18-FDG PET assay was used for chemotherapeutic efficacy based on rapid sodium and FDG uptake changes, immunohistological characterization.

Methods: PC-3 cells were injected to propagate prostate tumors in nude mouse. Gradient-echo images (3-D; 24 slices) were acquired on an Oxford 9.4-T MR microimager employing a double tuned proton-Na probe (ID 30 mm) and high-strength gradient insert coil (30 mT/m). Simultaneously, 18-FDG uptake was measured in animals using microPET with 50 mm bore. Following control tumor image acquisition, Taxotere (40 mg/kg; n = 9) was administered intravenously.

Results: After 24 hr, Taxotere induced a 30% (p <.001) increase in IR sodium image intensity and 55% (p <.005) enhanced 18-FDG uptake. Apoptosis, PET, and IR-Na changes were seen within hours of exposure of tumor to antineoplastic agents. Signal was suppressed arising from extracellular Na (primarily unbound) with long T1 without use of shift reagents. Histological sections (Feulgen labeling to nuclear DNA content by CAS 200 method) showed DNA content of cell nuclei, which varied according to their cell cycle (G1/S-G2M) and it was plotted as histogram. Various features indicative of drug effect and neoplasia could be identified, i.e., mitosis, apoptosis, and nuclear DNA content at different locations of tumor. Pentachrome staining offered “EC/IC volume” as marker of viable cells and cells at risk or dying cells. Grossly, viable cells showed average “EC/IC” 60–70% and higher mitotic index. Dying cells showed “EC/IC” 70% and more Apoptotic Index for the cells undergoing the process of advanced apoptosis. These post-Taxotere-injected tumor tissues exhibited decreased tumor viability seven to nine times, that is, mitotic figures from ~240 to ~15 mm² (in 80% tissue sections) and confirmed by lower proliferation index (PI), compared with control tumor tissues. Mitotic suppression correlated with increased IR image intensity for individual tumors (p <.002). Cell cycle of tumor cells showed reduced deploidy or aneuploidy in dividing tumor cells undergoing cell division in >80% tumor tissues after Taxotere treatment. Twenty-four hours post-Taxotere, injected animals showed single cell necrosis, possibly chemotherapeutic effect. Important features in coregistered histological and IC-Na, PET images were identified, which differed by <5% (histology vs. Na-MR images) and <7% (histology and PET images). On MRI images, necrosis and fluid were dark while active apoptosis had elevated [Na]_i signal intensity. PET images exhibited only hypermetabolic tumor regions matched with necrosis- and apotosis-rich regions.

Conclusion: In vivo sodium MRI may offer rapid prostate tumor cytomorphological characterization while PET may offer tumor hypermetabolic characterization in different regions. Dynamic changes in tumor cells after drug treatment can be monitored by MRI/PET method.

Abstract ID: 489

Computer-Assisted Quantitative Image Analysis of Cell Proliferation as Marker of Cancer Development

Frej Stenback, Veera Nayha, Jaakko Laitakari

Department of Pathology, University of Oulu, Oulu, Finland

The applicability of automated image analysis of immunohistochemical markers as indicators of development and progression of tumors in the human larynx was studied in order to increase our understanding of the process and possibly derive criteria of clinical significance. Amount, location, size, shape, and intensity of staining of proliferating cell nuclear (PCNA) antigen in precursors and in squamous cell carcinoma of the larynx were determined by computer-assisted morphometry with a sensitivity, accuracy, and reproducibility exceeding 99% in 12,538 cells altogether. Quantitative densitometry was performed using the CAS200 (Becton-Dickinson Cellular Imaging Systems, Leiden, Netherlands) automated image analyzer and proprietary software. The CCD camera was attached to a Reichert–Jung microscope, yielding appropriate linearity between light and dark settings. The measurements were done using 500-and 620 nm wavelengths. One field of measurement comprised 65.536 pixels; each pixel was measured individually. The results showed total PCNA expression to increase consistently during neoplasm formation. Individual preneoplastic cells in epithelial dysplasia expressed PCNA staining increasing with increasing cell size and antibody-staining intensity, in relation to increased degree of nuclear atypia. In malignant tumors, nuclear size decreased with decreasing differentiation, while antibody-staining intensity remained unchanged. The increased alterations in nuclear shape and in percent PCNA positive nuclei observed in dysplastic epithelium and squamous cell carcinomas were statistically significant using Spearman's correlation coefficient. In this study, identification of neoplastic transformation, separation of preneoplastic conditions, such as dysplasia from carcinoma, was best accomplished by analyzing nuclear shape. Squamous cell carcinomas consisted of two tumor cell populations with different nuclear shape and staining intensity. Automated image analysis of immunohistochemical PCNA immunolabeling is a relatively simple and rapid method of evaluating the proliferative activity of tumor tissue, but requires thorough standardization of laboratory methodology. Morphometric methods provided detailed information on large numbers of cells, useful for studies of tumor behavior and with potential clinical applications.

Abstract ID: 490

Imaging Gene Transfer Distribution and Efficiency in Lung

Marco Cecchini¹, Robert Malone², Bartosz Kubisa³, Antoinette Wetterwald¹, Ruth McDonald⁴, Erick Ayuni³, Jill Malone², Ralph Schmid³

¹Gene Therapy Laboratory, Department of Clinical Research and Department of Urology, University of Bern, Bern Switzerland, ²Gene Delivery Alliance/MIST Institute, Rockville, MD USA, ³Division of General Thoracic Surgery, University Hospital, Bern, Switzerland, and ⁴California Regional Primate Research Center and Center for Comparative Respiratory Biology and Medicine, University of California, Davis, CA, USA

As gene delivery and gene therapy research have progressed from ex vivo cell-based strategies to in vivo direct and targeted modification of cells and tissues, new types of reporter gene assays have been required. Increasing emphasis on the direct administration of gene vector preparations requires development of methods for analysis of the distribution and duration of transgene expression in whole organs and animals. Bioluminescent reporter imaging (BRI) is an ideal method for performing such analyses, but requires further adaptation for lung gene transfer research. For these studies, Fischer rat lungs were transfected with an efficient CMV plasmid-based transfection system encoding the P. pyralis luciferase. Intratracheal instillation of luciferin in PBS enabled direct detection of transgene expression in the transfected lung of the living rat. To provide more detailed information concerning the distribution of expression in the whole lung, a 37°C mixture of LMP agarose and luciferin was instilled intratracheally, the preparation was chilled at 4°C, and then was warmed to room temperature. The resulting inflated lung preparation was then equilibrated with a room temperature water bath, and semiquantitative BRI was performed. Intact lung preparations were directly imaged, and either airway microdissection or cross-sectional “breadloafing” was performed prior to BRI. For semiquantitative analysis of transgene expression, BRI signals from tissue samples were plotted as a function of sample luciferase content as determined by a standard luminometer assay. CMV-Luciferase plasmid was administered to the lungs of rats, providing for varying levels of luciferase expression. Alternatively, varying numbers of stably transfected, luciferase expressing cells were instilled into nontransfected rat lungs. Lungs were sectioned and analyzed by BRI using the agar technique. Both sides of each section were imaged and results were averaged to correct for differences in the depth of luciferase expressing cells. After imaging, the tissue sections were homogenized and assayed for luciferase protein using a luminometer assay and protein standard, thereby enabling correlation of BRI and luminometer readings. This method for BRI provided information concerning the overall distribution and quantity of luciferase expression within the lungs, but did not enable detailed cell analysis. We therefore transfected rat lungs with a CMV-eGFP plasmid, prepared the lungs with agarose as before but without luciferin, and imaged the unfixed tissue under fluorescence microscopy. For histopathologic analysis of transfected tissue, agarose-inflated lung sections were fixed in formalin, embedded in paraffin, and H&E stained, resulting in excellent quality sections of inflated lung tissue. BRI combined with these modifications provides a rapid and convenient method for semiquantitative analysis of the efficiency and distribution of tissue transfection.

Abstract ID: 492

Genomic Analysis of Squamous Cell Carcinoma Tumors Models Based on Spatial and Temporal Changes in Contrast-Enhanced MRI

Yi-Shan Yang, Samira Guccione, Gunyi Shi, Mark Bednarski

Department of Radiology, Stanford University School of Medicine, Stanford, California, USA

Gene and protein expression profiles in tumors can vary with temporal progression and spatial distribution. MRI is a powerful tool for the diagnosis of a variety of malignancies and is capable of visualizing changes in tissue contrast with micron spatial resolution and millisecond temporal resolution. We hypothesize that MRI can be used to determine temporal changes in tumor progression and reveal spatial heterogeneity within the tumor at different stages of growth for genomic analysis. Combination of imaging and microarray analysis can be used to validate potential molecular targets for new molecular imaging agents useful for cancer diagnosis and therapy. C3H/Km mice were transplanted with murine squamous cell carcinoma VII (SCCVII) cells. For each experiment, 2 × 10⁵ cells in 0.05-ml Hanks solution were implanted subcutaneously into the left flanks of mice. Tumors were imaged sequentially starting at a diameter of ~6 mm. A clinical 3-T MR Scanner was used with a custom-designed quadrature coil. Standard clinical spin-echo (T1-wt) and fast spin-echo (T2-wt) scan protocols were used. Postcontrast T1-wt image was obtained after an injection of gadolinium-DTPA. MRI contrast enhancing and nonenhancing regions were spatially marked at different stages of tumor growth, and the tumor tissue was surgically harvested for genomic analysis. We have found four distinct stages of tumor progression in the SCCVII murine tumor model based on MRI. We observed differences in gene expression profiles between each of these four stages of tumor growth and have identified several interesting gene products for each stage that are potential candidates as molecular targets for cancer imaging and therapeutics. In Stage 1, tumors show homogeneous contrast enhancement on both the T1 and T2-weighted images. Biopsy of the periphery and center of the tumor reveals no significant change in gene expression profile between these tumor regions. Stage 2 has shown heterogeneous T1-weighted contrast enhancement with homogeneous T2-weighted images. Biopsy of the contrast-enhancing and nonenhancing regions shows 10 genes that have a >2-fold up-regulation. Interestingly, despite the difference in the gene expression profile between the two regions, no difference is observed histologically. Stage 3 tumors show distinct T2-changes and heterogeneous contrast pattern in T1-weighted images. Genes that are up- or down-regulated with >2-fold change are distinct from those in Stage 1 but closely follow Stage 2 expression pattern. In addition, there are sets of new genes up-regulated that are not found on either Stage 1 or Stage 2. These include heat shock proteins, profilins, and calcium-binding proteins. Stage 4 shows significant T2 changes throughout the tumor with a narrow rim enhancement in contrast-enhanced T1-wt images, which is consistent with severe necrosis. We conclude that tissue sampling of tumors for genomic analysis using MRI-guided biopsy is sensitive to spatial and temporal changes in tumors. Clinical imaging can therefore serve as a powerful tool for discerning regions of distinct patterns of gene expression.

Abstract ID: 493

A Novel MMP-7-Selective Fluorogenic Substrate Detects Matrix Metalloprotease (MMP) Activity in Human Colon Xenografts

J. Oliver McIntyre, Barbara Fingleton, K. Sam Wells, Lynn M. Matrisian

Departments of Cancer Biology and Molecular Physiology, Vanderbilt University School of Medicine, Nashville TN 37232

Matrix metalloprotease (MMP) expression and activity have been implicated in both tumor growth and metastasis in animal models. However, the levels of different MMP activities in tumors have not been well characterized due to lack of suitable quantitative substrates. We have now synthesized a novel type of fluorogenic substrate (FSM-7) that can be selectively cleaved by MMP-7 (matrilysin). FSM-7 was prepared by linking a fluorescein (FL)-peptide, which includes a matrilysin (MMP-7) selective motif, to a polymeric scaffold. In a set of FSM-7 preparations with differing peptide/polymer ratios, the FL fluorescence is enhanced by up to ~16-fold by treatment with MMP-7 with minimal change in the fluorescence of tetramethylrhodamine (TMR), incorporated into FSM-7 as an internal reference. An in vitro fluorescence assay (measuring the FL/TMR ratio) has been developed and reveals efficient Ca²⁺-dependent cleavage of FSM-7 by MMP-7 (turnover number, ~24 sec⁻¹) with ~100-fold and ~10-fold slower cleavage by MMP-2 and MMP-3, respectively. A control polymer, lacking the peptide component of FSM-7, showed no change in FL/TMR ratio upon treatment with MMPs. By fluorescence microscopy, frozen sections of human colon tumor (HCA-7) xenografts (grown subcutaneously in a mouse model) exhibit significant Ca²⁺-dependent increases in the FL/TMR ratio of FSM-7 that correlates with the expression of MMP-7 and suggests the presence of active MMP-7 in these tumors. [Supported in part by NIH RO1 CA60867, P20 CA86283, and P30 CA68485.]

Abstract ID: 494

In Vivo Efficacy of Kinase Inhibitor SU11248 Against PC-3M-LUC Prostate Cancer Line Monitored by Bioluminescence Imaging

Lesley J. Murray, Tinya Abrams, Theresa Ngai, Dirk Mendel, Julie M. Cherrington

SUGEN, South San Francisco, CA 94080

SU11248 is an oral angiogenesis inhibitor with potency against kinases KDR, PDGF-R, c-kit, and flt3.

In biochemical assays, SU11248 exhibited competitive inhibition of KDR, PDGFR, and FGFR kinase activities with K_i values of 0.009, 0.008, and 0.83 μM, respectively In cellular assays, SU11248 inhibited VEGF-dependent KDR or PDGF-dependent PDGFR phosphorylation with IC₅₀ values of ~10 nM. In vivo, SU11248 caused regression or stasis of various established human tumor xenografts, such as A375 (melanoma), A431 (epidermoid), Colo 205 (colon), NCI-460 (lung), SF763T (glioma), MDA-MB-431 (breast), and MV411 (AML) in a dose-dependent manner, with linear PK. Caliper measurement of tumor volume is limited by inclusion of both lytic and necrotic areas, so there was often histological evidence of tumor destruction, although the effect was determined as stasis. Xenogen, Alameda, CA has developed technology to image live tumor cells in animals by detection of bioluminescence using luciferaselabeled tumor cell lines. Male nu/nu mice bearing subcutaneous luciferaseexpressing PC-3M tumors were evaluated for tumor growth, comparing caliper measurement of tumor volume to photon count based on the number of viable tumor cells present. At Day 14 when tumors were ~200–300 mm³, anaesthetized mice underwent initial imaging, starting 15 min after 150 mg/kg ip d-luciferin administration. Photon emission was captured by the IVIS system, and was analyzed with a customized version of the Living Image software. Mice were treated with SU11248 at 40 mg/kg/day in citrate buffer or vehicle alone by oral gavage. Tumor size was further evaluated at Days 21, 25, and 43/46, comparing in vivo imaging (photon count) and traditional caliper measurement. The results of this study indicate that SU11248 caused a 75.4% reduction of the number of viable tumor cells (p = .006) at Day 43 as indicated by the reduction in photon emission, but only 59% reduction in tumor volume (p = .003) at Day 46 measured using calipers. Bioluminescence imaging may thus give us better sensitivity to the effects of compounds to decrease live tumor cells. Models of lung colonization, bone metastasis, as well as orthotopic models, are under development for future efficacy studies of SU11248 and other SUGEN kinase inhibitors. SU11248 has recently entered clinical development in a Phase I dose escalating study in solid cancers.

Abstract ID: 495

Subcellular Near-Infrared Fluorescence Imaging of Cancer Cells

Akira Nakayama, John V. Frangioni

Division of Hematology/Oncology, Beth Israel Deaconess Medical Center, Boston, MA 02215

Near-infrared (NIR) fluorescence holds promise as a sensitive imaging and detection modality for epithelial malignancies. At present, little is known about how the in vivo detection of cancer cells is affected by the number of bound/accumulated fluorophore molecules, the photoproperties of the fluorophore, and tissue autofluorescence. In this study, we present detailed cellular accumulation and subcellular localization data on two different heptamethine indocyanine NIR fluorophores, IR-786 and its derivative IRDye78. Using bladder cancer model systems, we show that varying numbers of IR-786 or IRDye78 molecules can be targeted precisely to either mitochondria, endoplasmic reticulum, or the plasma membrane, and the effect of each subcellular location on fluorophore quenching can be determined. Directly, we compare cell and whole animal autofluorescence in the visible and NIR regions of the spectrum, and determine the number of cancer cells that can be detected in vivo by NIR fluorescence as a function of the above imaging parameters. We show that as few as 1250 cells loaded with 4.9 fmol IR-786 per cell, 12,500 cells loaded with 200 amol IR-786 per cell, and 12,500 cells loaded with 40 amol IRDye78 per cell can be detected in an injection volume of 10 ml at a distance of 0.9 mm below the surface of the animal. These results should permit a priori predictions of how to achieve in vivo NIR fluorescence imaging of epithelial tumors with the next generation of targeted NIR fluorophores.

Abstract ID: 496

In Vitro PET: Correlation of ¹⁸F-FDG Uptake and Glut-1 Expression in Tumor Cell Lines

Albert A. Geldof, Carla F.M. Molthoff, Richard T. Versteegh, Don W.W. Newling, Gerrit J.J. Teule, Adriaan A. Lammertsma

Departments of Nuclear Medicine/PET Center and Urology, VU University Medical Center, 1007 MB Amsterdam, The Netherlands

Molecular imaging of tumor glucose metabolism provides a powerful, noninvasive tool for detection, diagnosis, and follow-up of treatment response in cancer. In the present study, ¹⁸Fluorodeoxyglucose (FDG) accumulation in vitro by cells, derived from a number of different tumors, is determined quantitatively and correlated to the differential expression of a panel of markers for glucose transport (Glut-1) and phosphorylation (hexokinases: HK I, II, and III). Cells from the following tumor lines have been used: R3327-MATLyLu rat prostate tumor; PC-3 human prostate tumor; PC-12 rat pheochromocytoma; SW 1573 human colorectal tumor; T-47D and ZR75-1 human breast carcinomas; and TT2609 human follicular thyroid cancer. Various numbers of cells were plated in 12-well culture plates and incubated for 4 hr in glucose-free culture medium. FDG was added to an activity level of 100 μCi/ml at calibration (scan) time. After 1 hr at 37°C, the cells were washed and the incorporated activity was measured using a Siemens ECAT HR+ PET scanner and was analyzed by drawing regions of interest (ROI). The expression of Glut-1 and HK I, II, and III in cultured cancer cells was determined immunohistochemically in cytospin preparations and correlated with FDG accumulation. Considerable variation in FDG-uptake was recorded between different cell lines. These differences could be correlated to variable Glut-1 expression (and to a lesser degree to differences in HK I) but not to variation in HK II or III. From this study, Glut-1 expression appears to be the most important determinant for FDG PET signal in tumor cells under standardized in vitro conditions.

Abstract ID: 497

Novel Gallium(III) Complexes as Potential PET Imaging Agents for Probing Multidrug Resistance (MDR1) P-glycoprotein (Pgp) Transport Activity: Pharmacological and Pharmacokinetic Analysis

V. Sharma, J. Dahlheimer, C. M. Pica, D. Piwnica-Worms

Molecular Imaging Center, Mallinckrodt Institute of Radiology, Washington University Medical School, St. Louis, MO 63110

Among the best characterized mediators of chemotherapeutic failure in cancer are transport mechanisms due to overexpression of the ATP-binding-cassette family of transporters. P-glycoprotein (Pgp; a plasma membrane 170-kDa protein), the product of the MDR1 gene, and the related human multidrug resistant-associated protein (MRP1; 190 kDa) have proven roles in drug resistance while family members MRP2–6 are under investigation. Functional imaging of MDR1 Pgp-mediated transport activity would enable noninvasive assessment of chemotherapeutic resistance as well as MDR1 gene therapy in vivo. Therefore, exploiting the quantitative capacity of PET, organic scaffolds capable of accommodating PET isotopes to generate stable radiopharmaceuticals recognized by Pgp, is desired. Schiff-base ligands and amine phenol ligands that possess an N₄O₂ donor core to coordinate metals are known to generate stable gallium(III) complexes. As hydrophobic cationic complexes, these agents resemble other MDR agents in their physicochemical properties. We have synthesized and structurally characterized several Schiff-base and amine-phenol gallium (III) complexes, and a lead MDR imaging compound has been identified. The crystal structure of this compound revealed octahedral geometry with four secondary amine nitrogen atoms and two phenolate oxygen atoms in the equatorial and axial positions, respectively. Radiolabeled 67Ga-(for SPECT) or 68Ga-complexes (for PET) have been obtained through transmetallation reactions, and were found to be >95% radiochemically pure. The lead compound, namely, Ga-ENBDMPI [(bis(3-ethoxy-2-hydroxy benzylidene)-N,N′-bis(2,2-di-methyl-3-aminopropyl)ethylene-diamine)gallium(III)] was stable in physiologic buffer at 37°C for 72 hr. Further analyses of this compound extracted from the heart, liver, and kidney of mice 60 min postinjection indicated no evidence of metabolism. Further biochemical characterization of our lead radiopharmaceutical and its cellular accumulation in the presence of various MDR reversal agents and other non-MDR drugs was performed. In KB3-1 (Pgp-) and KB8-5 (Pgp+) cells, the radiopharmaceutical demonstrated net cellular uptake profiles inversely proportional to Pgp expression. In KB8-5 cells, GF120918, PSC 833, and LY 335979, potent and selective MDR modulators, fully reversed tracer uptake. In contrast, cis-platin and methotrexate, non-MDR agents, showed no effect on cellular accumulation. Like 99mTc-sestamibi, 67Ga-ENBDMPI demonstrated modest cross-reactivity with MRP1. However, no significant evidence of cross-reactivity with MRP2–6 was observed with this radiotracer. To further demonstrate utility in vivo, quantitative pharmacokinetic analysis was performed in wild-type (WT) and mdr1a/1b(−/−) gene disrupted (KO) mice with 67Ga-ENBDMPI. Compared to WT control, the tracer demonstrated markedly increased initial penetration and 17-fold enhanced retention in brains of KO mice, consistent with permeation through the known Pgp-mediated transport barrier normally present in brain capillaries in vivo. The evidence suggests that 67/68Ga-ENBDMPI may be suitable to assess functional transport activity of MDR1 Pgp in vivo.

Abstract ID: 499

Magnetic Resonance Microimaging Technique for the Noninvasive Study of Brain Tumor Invasion

K. D. Bahk, P. N. Venkatasubramian, A. M. Wyrwicz

Center for MR Research, ENH Research Institute Departments of Biochemistry, Molecular and Cell Biology, Neurobiology and Physiology, Northwestern University

We have developed a magnetic resonance microimaging technique that allows the noninvasive study of brain tumor invasion in a true in vivo environment. This technique has the ability to differentiate between invading tumor cells and adjacent brain tissue. Our studies initially identified and optimized a method to label C6/LacZ glioma cells with the MR contrast agent gadopentetate dimeglumine in order to achieve this differentiation. Optimal labeling conditions via electroporation of C6/LacZ rat brain tumor cells were established to provide satisfactory balance between cell toxicity (determined by Trypan blue assay) and the maximal intracellular concentrations of gadolinium/cell as determined using T1-spectroscopic analysis. The intracellular presence of gadolinium decreases the T1-relaxation times of the intracellular water molecules, effectively labeling these cells. High concentrations of contrast agent within cells result in more prominent decreases in T1-relaxation times, which translate to greater differences in signal intensity as determined spectroscopically and observed in MR images. The effects of our labeling strategy were then evaluated in further detail. Labeling of cells via electroporation was determined to be homogenous within a cell population, as determined by electron probe microanalysis (EPM). Quantitation of gadolinium was achieved using inductively coupled atomic emission spectrometry (ICP-AES). In vitro evaluations of labeled cells indicated no deleterious effects on morphology or proliferative rate. T1-spectroscopic analysis illustrated significant differences in magnetic properties between labeled and nonlabeled cells. These differences, after labeled cells were stereotactically implanted into the caudate putamen of syngeneic Fisher 344 rats, were clearly visible in in vivo MR images as areas of hyperintensity, thus confirming that our labeling strategy allows the noninvasive visualization of brain tumor invasion. In vivo images were acquired on a 4.7-T Bruker/GE instrument. Areas of hyperintensity correlate well with LacZ-stained brain tissue slices of the identical tissue. Cell division was expected to decrease the intracellular concentration of gadolinium label, and therefore, the signal intensity observed in MRI. Through in vitro and in vivo studies, the effects of cell division on the intracellular gadolinium concentration were characterized, providing insight into the maximal number of days (10 days) during which the process of invasion can be visualized using this method. In conclusion, we have developed a noninvasive method to study brain tumor invasion in the previously unobservable initial growth stages. This strategy will facilitate greater understanding of brain tumor invasion in the ideal microenvironment.

Abstract ID: 500

Enhancement of Tumor Uptake of Gallium-67 by Photodegraded Nifedipine

Kathryn A. Morton¹, Jean-Baptiste Roullet²

¹Department of Radiology, Wake Forest University School of Medicine, Winston-Salem, NC, 27157 and ²Department of Surgery, Oregon Health Sciences University, Portland, OR, 97201

Rationale: All current tumor-imaging methods present limitations. Gallium-67 is a radiometal used for many years with variable success for tumor imaging. The goal of this research is to selectively increase tumor uptake of ⁶⁷Ga. Tumor uptake of ⁶⁷Ga has traditionally been thought to be transferrin-dependent. However, we previously reported that the transferrin-independent uptake of ⁶⁷Ga may be more important in tumors than normal tissues, and is greatly enhanced by light-exposed nifedipine, but not by the native compound. The active photo derivative of nifedipine was isolated and characterized. Selectivity of the compound in promoting uptake of ⁶⁷Ga by tumors was evaluated in a murine cancer model. The kinetics and mechanism of action were addressed in cells.

Methods: The photoderivatives of nifedipine were isolated, purified, physically characterized, and evaluated for effect on cellular uptake of Ga³⁺. The active derivative was tested for stability, protein binding, calcium channel blocking activity, and spectral interaction with Ga³⁺. The effect of the compound on ⁶⁷Ga uptake and distribution was assessed in tumor-bearing mice. The kinetics and mechanism of action were evaluated in cultured cells.

Results: The sole compound produced by fluorescent irradiation of an ethanolic solution of nifedipine, and the predominant product with UV irradiation, is the fully aromatic nitroso derivative, which we have named “nitrosipine.” Nitrosipine efficiently promotes cellular uptake of Ga³⁺. Nitrosipine is stable under conditions of laboratory storage and is unaffected by acidification and subsequent neutralization. It displays time-dependent protein binding. In mouse arteries, nitrosipine is a weak calcium channel blocker, compared to nifedipine. Gallium³⁺ induces a unique, dose-dependent shift of both the UV and ¹H NMR spectra of nitrosipine, suggesting that nitrosipine may interact directly with Ga³⁺. In a murine tumor model, nitrosipine selectively enhances uptake of ⁶⁷Ga in tumors in mice by 3–10 fold, while uptake by normal tissues and organs is reduced or unchanged. In vitro, the effect of nitrosipine on ⁶⁷Ga uptake is rapid in onset, reversible, and saturable. It is enhanced by low extracellular pH and also lowers intracellular pH, suggesting a proton-coupled transporter similar to the divalent metal transporter-1 (DMT-1), the transporter expressed in absorptive enterocytes, which transports Fe²⁺ and other metal ions, also homologous to NRAMP2. Sodium is not required, but potentiates the nitrosipine effect, suggesting the coordinated activity of an active Na⁺/H⁺ antiporter. Nitrosipine also promotes the cellular uptake of a number of other metal cations, including Fe²⁺, Fe³⁺, and Cu²⁺.

Conclusion: Nitrosipine is the prototype of a new class of drugs that promotes entry of isotopes of gallium and other metals into tumor cells, a property that may be exploited for tumor imaging, radiotherapy, or as an adjunct to chemotherapy. The selective action on tumors may result from the acidic microenvironment of tumors. Nitrosipine may activate a saturable proton-coupled transporter similar to the DMT-1 described for iron and other metal ions.

Abstract ID: 502

Correlation of Magnetic Resonance Imaging and Spectroscopy with Histopathologic Features of Grades II and III Glioma

T. McKnight¹, R. Henry¹, T. Love¹, D. Vigneron¹, M. Berger², M. McDermott², S. Nelson¹

Departments of ¹Radiology and ²Neurological Surgery, University of California, San Francisco, USA

Introduction: The heterogeneous nature of glioma presents a considerable challenge for surgeons attempting to choose a region to biopsy during tumor resection. Treatment effectiveness and prognostic accuracy are greatly improved when the most malignant region of the tumor is biopsied. This study is part of an ongoing investigation into the usefulness of preoperative MR spectroscopic imaging (MRSI) and diffusion tensor MRI for identifying high-grade regions within nonenhancing gliomas. Two primary histological features that are used to differentiate between Grade II and III lesions are cell density and proliferative rate. The apparent diffusion coefficient (ADC) that is obtained with diffusion tensor MRI is presumed to reflect tumor cellularity. Both cellularity and proliferation are thought to contribute to the MRS choline (Cho) peak. We have developed an index, the Cho-NAA Index (CNI), for interpreting 3-D-MRSI data from glioma patients (McKnight, et al. JMRI, 2001). However, the relationship between the histopathologic and MR features of the individual biopsies has not been investigated. The maximum value of the CNI was found to be higher in patients with Grade III glioma as compared with Grade II, suggesting that it may be a useful indicator of tumor aggressiveness. The current study compares the preoperative CNI with the proliferative index and cellularity of excised tissue samples.

Methods: Twenty patients (12 Grade II and 8 Grade III) underwent 3-D-MRSI prior to surgery. During the tumor resection, the locations of one to three biopsies were recorded on MRIs that were subsequently used to determine the preoperative CNIs and ADC at the same locations. A total of 33 paraffin-embedded specimens were collected and clinically classified as Grade II (n = 20) or Grade III (n = 13) glioma. Immunohistochemical MIB-1 labeling was performed on additional sections from the biopsies. The cellularity was determined by dividing the total number of cells counted on each slide by the number of microscopic fields.

Results: The CNI correlated well with both the MIB LI (p <.01) and the cellularity (p <.001) of the corresponding tissue sample. Within the Grade III samples alone, there was also a significant correlation with MIB LI(p <.05) and cellularity(p <.02). However, we found no significant correlations among the Grade II tumors.

Discussion: The CNI was strongly correlated with both the proliferative rate and the cellularity of untreated Grades II and III glioma. The absence of a correlation with either histologic parameter among the Grade II tumors suggests that the observed correlation directly reflects the more malignant features of the higher-grade tumors. This finding supports the use of CNIs from preoperative MRSI data for detecting malignant foci within low-grade lesions. We will perform the same correlations with ADC values at the biopsy locations to determine whether the addition of diffusion-weighted MRI improves the accuracy of detecting high-grade regions in low-grade tumors.

Abstract ID: 503

Noninvasive Optical Imaging of Epithelial Tissue

Judith R. Mourant, Tamra Thayer, Tamara M. Johnson, Susan Carpenter, James P. Freyer

The goal of this work is to develop an imaging method based on light scattering that will noninvasively measure morphological characteristics of epithelial tissue. Noninvasive, polarized, light scattering images of tissue can provide information on microscopic tissue structure. For example, the interaction of light with cultured cells has been shown to depend on the growth rate of cells, and the structures causing light scattering have been shown to be on the order of tenths of microns. We will present the scientific basis for using polarized light to noninvasively measure microscopic tissue properties. For example, the results of angularly resolved, light scattering measurements of epithelial cells will be used to demonstrate the type of information that is available and to motivate data analysis of optical images. Polarized, optical images of scattering media, such as tissue, are obtained by illuminating a point and then imaging around that point. The area imaged is a few millimeters in radius. Images of physical tissue phantoms demonstrating a sensitivity to small changes in the microscopic structure will be presented. Calculation of the expected light penetration into tissue will also be presented. Finally, we plan to present images of dense suspensions of mammalian cells.

Abstract ID: 505

Human Thymidine Kinase Type 2—A Novel Nonimmunogenic Reporter Gene for Noninvasive Imaging in Humans

V. Ponomarev, M. Doubrovin, I. Serganova, L. Ageyeva, T. Beresten, S. Sagomonyan, J. Balatoni, R. Finn, R. Blasberg, J. Gelovani Tjuvajev

Memorial Sloan Kettering Cancer Center, New York, NY

Objective: Develop a novel nonimmunogenic human reporter gene for imaging different molecular-genetic processes using nuclear imaging techniques (PET).

Methods: Human thymidine kinase type 2 (TK2) cDNA, truncated at the N-terminal end, was fused with green fluorescent protein (GFP) cDNA to generate a fusion protein hTK2GFP to facilitate selection of the transduced cells. The resulting hTK2GFP cDNA (fusion gene) was cloned into an MoMLV-based retrovirus and was used to transduce the human glioma U87 cell line. A mixed population of transduced U87/hTK2GFP+ cells was obtained by selection with FACS, and the level of hTK2GFP enzymatic activity was assessed in vitro using the [¹⁴C]FIAU-[³H]TdR radiotracer accumulation assay. Two xenografts were generated from U87/hTK2GFP+ cells and wild-type U87 cells in opposite shoulders of SCID mice. When the xenografts reached palpable size (~0.5 cm in diameter), [¹²⁴I]FIAU (~150 μCi/animal) was injected intravenously and microPET imaging was performed at 2 and 24 hr later.

Results: The hTK2GFP-transduced cells had high-intensity green fluorescence on fluorescent microscopic and FACS analysis (mean fluorescence of the sorted population was 460). The K_i of [¹⁴C]FIAU in U87/hTK2GFP+ cells was 0.068 ± 0.020 (SD) ml/min/g, whereas K_i for wild-type U87 cells was 0.007 ± 0.002 ml/min/g; the transduced to nontransduced ratio was about 10. In vivo PET imaging demonstrated localization of radioactivity only in the U87/hTK2GFP+ transduced xenograft, with background levels of activity in the wild-type, control xenograft. Tissue sampling [¹²⁴I]FIAU radioactivity assays at 24 hr postinjection yielded values of 0.24 ± 0.09 and 0.038 ± 0.002 %dose/g in U87/hTK2GFP+ and U87 xenografts, respectively. U87/hTK2GFP xenograft/plasma and xenograft/muscle ratios were 87 ± 17 and 27 ± 11, respectively; the U87 xenograft/plasma and xenograft/muscle ratios were 14 ± 2 and 4.0 ± 0.3, respectively.

Conclusion: We have tested a new hTK2 reporter gene for potential PET imaging in human subjects with [¹²⁴I]FIAU. The hTK2 reporter gene is likely to be nonimmunogenic and will allow for more widespread monitoring of different molecular-genetic processes in human subjects. In this report, we have used the dual hTK2GFP reporter gene to demonstrate imaging feasibility. [Supported by NIH grants CA69769, CA86438, CA57599, CA76117, CA98023 and DOE grant FG03-86ER60407.]

Abstract ID: 506

Time-Lapse Video of GFP Actin Binding Proteins Shows Simultaneous Stretching and Contraction of Stress Fibers, In Vivo

Lynda Peterson¹, Zenon Rajfur¹, Amy Shaub¹, Magnus Edlund², Carol Otey³, Ken Jacobson^1,4, Keith Burridge^1,4

¹Department of Cell and Developmental Biology, ³Department of Molecular and Cellular Physiology, ⁴Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC 27599, and ³Department of Urology, University of Virginia, Charlottesville, VA 22908

Cell migration and adhesion to the underlying substrate is critical for normal cell functions such as embryonic development and wound healing, as well as tumor cell metastasis and tissue invasion. Both migration and adhesion are dependent in part on forces generated by actin–myosin II-based contraction. Stress fibers are bundles of filamentous actin, the assembly of which depends on acto-myosin contractility. These structures are anchored at transmembrane adhesive complexes that differ in their mechanical properties, protein composition, and regulatory effectors depending upon whether they occur in the protrusive zone behind the leading edge or in the retractin tail of migrating cells. Using a cell model in which we can control the rate of cell retraction, we examined contractility dynamics by time-lapse video microscopy of green fluorescent proteins (and variants) fused to stress fiber proteins and expressed in fibroblasts and gerbil fibroma cells. We find that unlike muscle fibers, stress fibers do not display uniform contraction along their entire length. Simultaneous stretching of central regions and contraction of peripheral zones nearest adhesive structures suggest localized composition and regulation of stress fiber components. We are currently pursuing chromophore assisted laser inactivation (CALI) to examine the effects of localized inactivation of stress fiber proteins as well as those associated with adhesive complexes.

Abstract ID: 507

Better Gene Expression with Replication Competent Versus Defective Adenoviruses: Confirmation Using the Sodium Iodide Symporter Gene

K. N. Barton¹, S. L. Brown¹, S. Jhiang², D. Tyson¹, J. H. Kim¹, S. O. Freytag¹

¹Radiation Oncology, Henry Ford Hospital, Detroit, MI 48202 and ²Department of Physiology and Cell Biology, Ohio State University, Columbus, OH 43210

A major limitation of gene therapy is the inability to monitor vector biodistribution and persistence and therapeutic gene expression in vivo. Noninvasive techniques that allow for tracking of the spatial and temporal distribution of the vector and its gene products are desirable over invasive techniques that are often painful and inconclusive. One way to monitor vector biodistribution and persistence in vivo is to express a gene whose product can be followed noninvasively. The sodium iodide symporter (NIS) is expressed mainly in the thyroid gland, stomach, and kidneys and transports ions such as iodide (¹²³I) and technetium (^99mTc). Radiolabeled iodide or ^99mTc are commonly used in the clinic to generate real-time whole body scans in conjunction with a gamma camera. We have generated a replication-competent adenovirus (Ad5-yCD/TK-hNISrep) that contains a therapeutic yeast cytosine deaminase (yCD)/herpes simplex virus thymidine kinase (HSV-1 TK) fusion gene in the E1 region and the human NIS gene in the E3 region. Expression of hNIS in cells infected with Ad5-yCD/TK-hNISrep in vitro was confirmed by Western blotting and immunofluorescence with an anti-hNIS antibody. DU145 prostate adenocarcinoma cells infected with Ad5-yCD/TK-hNISrep transport ¹²³I and ^99mTc in a perchlorate-sensitive manner were 1000-fold more efficient than cells infected with a replication-defective adenovirus expressing hNIS (Ad5-hNIS). hNIS-dependent uptake of ^99mTc was readily detected noninvasively in experimental tumors injected with 3 × 10¹⁰ vp Ad5-yCDTK-hNISrep. hNIS-dependent uptake of ^99mTc was also detected noninvasively in the liver 14 days postviral injection in mice injected systemically with 1 × 10¹⁰ vp Ad5-yCD/TK-hNISrep. To monitor the spatial distribution and persistence of the vector in a large animal model that resembles the human situation, hNIS-containing adenoviruses were injected into the dog prostate (1 × 10¹⁰ vp Ad5-hNIS into the left lobe and 1 × 10¹⁰ vp Ad5-yCD/TK-hNISrep into the right lobe). Two days later, animals were administered 40 mCi Tc and the prostate was removed and imaged with a clinical gamma camera. The spatial distribution of each virus within the prostate gland was readily detected. Furthermore, the Ad5-yCD/TK-hNISrep-infected prostate cells in the right lobe took up more Tc than cells of the left lobe infected with Ad5-hNIS. Studies are in progress that will determine the biodistribution and persistence of replication-competent versus replication-defective hNIS-containing adenoviruses following intraprostatic and systemic administration. The data suggest that hNIS can be used to monitor noninvasively the biodistribution and persistence of adenoviral gene therapy vectors in humans.

Abstract ID: 508

Bcl-2 mRNA as a Target for Antisense Imaging Agents: Digital Imaging Fluorescence Microscope Studies of F-G3139, a Fluorescently Labeled Phosphorothioate

D. E. Callahan¹, B. Parvin², S. E. Taylor¹

¹Life Sciences Division and ²National Energy Research Scientific Computing Center, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720

We have studied the uptake, intracellular distribution, retention, and biological effects of F-G3139 in cell lines containing constitutively high (MCF-7 “WTC”) and low (MCF-7 “ADR”) levels of bcl-2 mRNA. F-G3139 is 5′-(FITC)-sTsCs TsCsCsCsAsGsCsGsTsGsCsGsCsCsAsT-3′, where a fluorescein phosphoramidite with a hexyl linkage was used to attach FITC to the 5′ end of the all-phosphorothioate G3139 oligonucleotide. G3139 is a negatively charged 18mer that targets the first six codons of the bcl-2 mRNA open reading frame. Living cells were imaged (63 × oil-immersion objective, NA 1.3) for several hours in a temperature-controlled microscope perfusion chamber using both transmitted and fluorescence light. A Hamamatsu C4880 cooled CCD camera was used to acquire digital images. Washout curves generated from WTC and ADR cells loaded with F-G3139 were significantly different at 37°C, with F-G3139 washing out less readily from the high bcl-2 WTC cells. However, ADR cells have higher levels of the transmembrane multidrug transporter molecule P-glycoprotein (Pgp) than WTC cells, thus, the presence of Pgp and not the intrinsically lower levels of bcl-2 mRNA may be responsible for the lower accumulation and retention in ADR cells. In order to determine whether F-G3139 is a substrate for Pgp, we obtained uptake and washout curves for F-G3139 in the presence and absence of 25 μM verapamil, a competitive inhibitor of Pgp; preliminary results indicate that F-G3139 may be a substrate for Pgp. In a separate set of studies, living WTC and ADR cells were prestained with the membrane-permeable nuclear dye Hoechst 33342 (“H42,” 1.75 μM) under serum-free conditions (DPBS containing glucose and pyruvate, pH 7.4). This fluorescent dye is often used to stain the nucleus of living cells. We found that cells exposed to H42 for several hours rounded up and underwent a membrane permeability transition characteristic of the early stages of apoptosis. This increase in membrane permeability was detected using the nuclear stains YO-PRO-1 and/or YO-PRO-3. In rounded cells that stained positive for increased membrane permeability, a large increase in the amount of F-G3139 fluorescent signal was also observed. Both WTC and ADR cells exhibited a similar increase in fluorescence intensity that was concentrated in the nucleolar regions of the nucleus. Under similar conditions, in the absence of H42, cells of both types that remained flattened and well attached displayed a distinctive F-G3139 cytoplasmic staining. In a third set of studies, apoptosis was induced in WTC and ADR cells by glucose deprivation in the presence of glucose-free fetal bovine serum. Under these conditions, exposure to 0.7, 2.0, and 4.0 μM F-G3139 was cytotoxic to WTC, but not ADR cells. This is consistent with previous results which have demonstrated that nonfluorescent anti-bcl-2 antisense compounds are cytotoxic to cells with constitutively high levels of bcl-2. [Funded through the Office of Biological and Energy Research, Molecular Nuclear Medicine Program, in the Office of Science, U.S. Department of Energy, under contract DE-ACO3-76SF00098.]

Abstract ID: 510

PET Imaging of TGFβ Signal Transduction Pathway Activity

I. Serganova, V. Ponomarev, M. Doubrovin, L. Ageyeva, T. Beresten, S. Soghomonian, J. Balatoni, R. Finn, R. Blasberg, J. Gelovani

Tjuvajev Memorial Sloan Kettering Cancer Center, New York, NY

Objectives: Reporter gene imaging is an indirect approach to visualize transcriptional and posttranscriptional regulation of target gene expression. The cytokine TGFβ plays a dual role in tumorigenesis. It stimulates the proliferation of mesenchymal cells while inhibiting the growth of most normal epithelial cells. To image the activity of the TGFβ signal transduction pathway activity during tumorigenesis and anticancer treatment, we generated a cis-TGFβ/TKGFP reporter vector.

Methods: Double-stranded complementary oligonucleotides (containing sequences from the mouse germline IgA promoter) were used to construct IgA-Smad-AML-TKGFP reporter gene, which contains both AML1 and Smad consensus binding sites. The DNA fragment with the AML1 and Smad binding sites was subcloned into the EcoRI and XbaI sites of the dxNFAT/TKGFP-Neo vector in place of the NFAT enhancer element (Ponomarev et al., 2001). Thus, the HSV1-tk/GFP reporter gene was linked to the enhancer elements specific for Smad/AML1 transcription factors. The resulting plasmid was transfected into the H29 transient retroviral vector producer cells. MDA-MB-231 breast cancer cells were transfected with the cis-TGFβ/TKGFP vector. Stable transfectants were selected with G418. MDA-MB-231 cells containing the IgA-Smad-AML-TKGFP reporter system were exposed to the various concentrations of the human TGFβ1 for 48 hr with and without starvation. The level of TKGFP expression was assessed by fluorescence microscopy and quantified by FACS and a radiotracer assay ([¹⁴C]FIAU and [³H]TdR).

Results: A plateau of transcriptional activation of the TGFβ signaling pathway was observed after 16 hr of exposure to minimal concentrations of TGFβ1 (2 ng/ml) and starvation. TGFβ1-treated cells exhibited high-level TKGFP reporter protein as measured by FACS (>10² relative fluorescence units). Significantly higher levels of the FIAU/TdR ratio (0.10 ± 0.02) were observed under these conditions as compared to baseline, nonstimulated cells (0.007 ± 0.0001).

Conclusions: The level of FIAU accumulation measured in stimulated IgA-Smad-AML-TKGFP transduced cells indicates that successful imaging of TGFβ-induced transcriptional activity with [¹²⁴I]FIAU and PET is feasible. This would be useful in assessing novel antitumor drugs that act through inhibition of the TGFβ signaling pathway. [Supported by NIH grants CA69769, CA86438, CA57599, CA76117, CA98023 and DOE grant FG03-86ER60407.]

Abstract ID: 511

Imaging Mammary Tumors in the APCmin+/Mouse Model with NM404-A Second Generation Phospholipid Ether Analog

J. P. Weichert¹, A. R. Moser², M. A. Longino³, R. E. Counsell³

Departments of ¹Radiology, ²Human Oncology, University of Wisconsin, and ³Department of Pharmacology, University of Michigan

Our strategy for the site-specific delivery of imaging agents is based on a biochemical approach whereby naturally occurring compounds known to be stored or metabolized in selective tissues serve as carriers for the radiologic moiety. Based on a literature report that a variety of animal and human tumors contained much higher concentrations of naturally occurring ether lipids in their cell membranes than normal tissues, we synthesized a variety of radioiodinated phospholipid ether analogs (PLE) as potential tumor imaging agents. Our prevailing hypothesis is that phospholipid ethers become trapped in tumor membranes because of their inability to become metabolized by membrane-associated phospholipases PLA, PLC, and PLD, and cleared. These radioiodinated PLE analogs have displayed striking tumor avidity in all nine xenograft animal tumor models examined and have recently exhibited tumor avidity in the spontaneous ApcMin/+ mouse mammary carcinoma model as well. Moreover, extraction of tumors following administration of radioiodinated phospholipid ethers showed the presence of only the intact agent, whereas analysis of the urine and feces revealed only metabolites. It is the differential clearance rates of phospholipid ethers from normal cells versus tumor cells that form the basis of this concept. Extensive structure–activity relationship studies resulted in the synthesis, radiolabeling, and evaluation of over 20 radioiodinated PLE as potential tumor-selective imaging agents. It was found that any chemical modification of the phosphocholine moiety or shortening the chain length of the iodophenylalkyl moiety to less than eight methylenes resulted in little or no tumor uptake. The most promising of the original series of compounds, NM324, was evaluated in several human lung cancer patients. The agent localized in lung lesions, but image quality was compromised by high background liver radioactivity. A second series of compounds was made in order to enhance plasma half-life and to decrease background liver activity. In subsequent comparative tissue distribution and imaging studies conducted in several different tumor models, a second generation compound, namely, NM404, was identified that afforded better tumor imaging characteristics than its predecessor, NM324. NM404 recently passed toxicity at 200 times the anticipated imaging dose and received IND approval for diagnostic use in a human prostate cancer patient trial. These agents become trapped in the outer membranes of viable tumor cells where they are retained for prolonged periods. Current studies are underway in the ApcMin/+ mouse mammary carcinoma model to determine whether 125I-NM404 can distinguish between preneoplastic and neoplastic lesions. Initial imaging results have shown striking uptake and prolonged retention in neoplastic lesions. Dosimetry studies are also being conducted to evaluate the appropriateness of these agents as radiotherapy agents due to their exceedingly long tumor retention properties.

Abstract ID: 512

Direct Comparison of FIAU and FHBG as HSV1-tk PET Imaging Probes

J. Gelovani Tjuvajev¹, M. Doubrovin¹, T. Akhurst¹, S. Cai¹, J. Balatoni¹, M. Alauddin², R. Finn¹, B. Beattie¹, S. Larson¹, P. Conti², R. Blasberg¹

¹Memorial Sloan Kettering Cancer Center, New York, NY and ²University of Southern California, Los Angeles, CA

Objectives: To compare the efficacy of two currently available radiolabeled probes for imaging HSV1-tk expression in vivo with PET.

Methods: [¹⁸F]FHBG and [¹²⁴I]FIAU were prepared as previously described. Two established cell lines, stably transduced RG2TK+ and wild-type RG2, were used for paired comparisons of probe accumulation in vitro and for paired comparisons of subcutaneous xenografts produced from these cell lines in athymic rnu/rnu rats. The in vitro experiments were run over 180 min and the incubation medium contained [¹⁸F]FHBG and [¹⁴C]FIAU plus [³H]TdR [Tjuvajev et al., Cancer Res 55: 6126, 1995]. The in vivo experiments involved [¹²⁴I]FIAU imaging and/or tissue sampling at either 2 or 24 hr; for [¹⁸F]FHBG, imaging and tissue sampling were performed at 2 hr.

Results: The in vitro paired probe uptake comparisons in RG2TK+ cells, normalized to TdR, indicates that FIAU accumulation was 15-fold greater than FHBG. The calculated K_i values were 0.022 ± 0.001 (SD), and 0.317 ± 0.066 ml/min/g cells for FHBG and FIAU, respectively. The 2-hr in vivo studies yielded similar results: the 2-hr RG2TK+ xenograft accumulation was 0.074 ± 0.049 and 1.22 ± 0.21 %dose/g tissue for FHBG and FIAU, respectively. The RG2TK+/RG2 ratio for FHBG and FIAU was 5.9 ± 2.3 and 11.3 ± 2.6, and the RG2TK+/muscle ratio was 4.9 ± 2.2 and 18.1 ± 8.8, respectively. At 24 hr after FIAU administration, the RG2TK+ xenograft and ratio values were 1.53 ± 0.40 %dose/g, 65 ± 21, and 124 ± 58, respectively. The accumulation profiles of FIAU and FHBG in RG2TK+ and RG2 xenografts demonstrated that FIAU rapidly accumulates and reaches a plateau in RG2TK+ xenografts, whereas FIAU is being cleared from the RG2 xenografts during the 2-hr period of imaging. In contrast, FHBG is being cleared from both TK-transduced and wild-type xenografts over the 2-hr imaging period. FHBG clearance from RG2TK+ xenografts is slower than that in RG2, and appears to approach a plateau.

Conclusions: The in vitro uptake studies and in vivo PET imaging results demonstrate that FIAU is a substantially better probe for imaging HSV1-tk expression at 2 and at 24 hr, with greater sensitivity, dynamic range, and lower abdominal background radioactivity than FHBG. [Supported by NIH grants CA69769, CA86438, CA57599, CA76117, CA98023 and DOE grant FG03-86ER60407.]

Abstract ID: 514

Visualization of Adoptively Transferred Immune T Lymphocytes at the Tumor Site Using Positron Emission Tomography

Purnima Dubey¹, Helen Su², Nona Adonai³, Shouying Du¹, Jonathan Braun⁴, Michael Phelps⁵, Sanjiv Gambhir⁵, Owen Witte^1,6

¹Department of Microbiology, Immunology and Molecular Genetics, ²Molecular Biology Institute, ³Department of Molecular and Medical Pharmacology, ⁴Department of Pathology, ⁵Crump Institute for Molecualr Imaging, and ⁶Howard Hughes Medical Institute, University of California, Los Angeles, CA

To assess more accurately the progress of the antitumor immune response over time, it is critical to develop new and more sensitive methods for repeated and noninvasive measurement of T cell migration and function. The methodology can then be directly translated to use in the clinics for repeated monitoring of the patient to determine the efficacy of adoptively transferred T cells in the destruction of cancer. In this work, we evaluate the use of positron emission tomography (PET) as a tool to repeatedly and noninvasively image the migration and localization of immune T cell to the tumor site. In order to mount an effective antitumor response, the cellular arm of the immune system must be engaged. The effectiveness of T cells in antitumor responses is generally monitored by indirect methods, such as measurement of tumor volume, macroscopic tumor progression or regression, detection of cytolytic activity in vitro. In order to monitor the immune response over a span of time, it is necessary to sacrifice multiple animals for each time point. As a proof of principle, we have used a virally induced murine rhabdomyosarcoma cancer model to optimize methods for marking tumor-specific CTLs with PET reporter genes as well as technical aspects of imaging the tumors and T cells. Immunocompetent mice are challenged with the MSV/MLV virus complex, which is rejected by a strong CTL response. Immune splenocytes are isolated, and T cells marked with the sr39TK PET reporter gene via retroviral infection. Sr39TK-marked T cells are then adoptively transferred into immunodeficient mice bearing either an MSV/MLV induced tumor, or control tumor. Three days after T cell transfer, animals are injected with ¹⁸F-FHBG, the radioactive substrate for sr39TK. The isotope is then detected by PET. A clear and strong signal generated by the isotope is detected at the site of the antigen-positive tumor. Immunohistochemical detection of the T cells using α-CD3 staining shows a large number of T cells in the antigen-positive tumor, as compared to the antigen-negative tumor. The same animals are repeatedly imaged several times over a period of 10 days. In addition, we show that MSV/MLV-immune T cells migrate to the site of MSV/MLV-induced tumor and emit a much stronger signal than adoptively transferred naïve sr39TK-marked T cells. PET is a sensitive method that can be used to track the migration and localization of antigen-specific T cells.

Abstract ID: 515

Direct Imaging of Camptothecin Anticancer Drugs in Optically Dense Media Using Fluorescence with Multiphoton Excitation

Thomas G. Burke, Ignacy Gryczynski, Zygmunt Gryczynski, Henryk Malak, Joseph R. Lakowicz, Vincent Wallace, Bruce J. Tromberg

The family of camptothecin anticancer drugs exhibits a high intrinsic fluorescence. In order to detect these agents directly in optically dense samples such as blood and tissue, we have employed fluorescence spectroscopic and microscopic methods using multiphoton excitation. We have demonstrated that topotecan and other camptothecins are capable of being detected in human plasma and whole blood using two-photon excitation at 730 or 820 nm. These wavelengths are longer than the main absorption bands of hemoglobin. Two-photon excitation of topotecan was demonstrated by a quadratic dependence of the emission intensity on the incident power, as compared to a linear dependence for one-photon excitation at 410 nm. In phosphate buffered saline containing human albumin, topotecan displayed a two-photon cross-section of 36 × 10⁻⁵⁰ cm⁴ sec/photon and an excited-state lifetime of 4.3 nsec. The observed emission centered at 525 nm was shown to be topotecan from the similarity of the emission spectrum and decay times observed for one-photon and two-photon excitation. Topotecan was detected at concentrations as low as 0.05 and 1 μM in plasma and whole blood, respectively. Cancer cells and tumor spheroids exposed to camptothecins were also imaged using a two-photon scanning fluorescence microscope system (TPM). The TPM system consisted of a 5-W Verdi laser (Coherent), which is used to pump a Mira 900F Ti:Sapphire laser (Coherent) that was used as the two-photon pulsed excitation source. The ultrafast Ti:Sapphire laser was tunable between 690 and 1000 nm, thus allowing for two-photon imaging of many UV excitable camptothecins. The two-photon excited fluorescence from the tissue was detected using a single-photon counting detection system that consisted of two photomultiplier tubes arranged perpendicularly, one optimized for green light, the other for red light, thus allowing for simultaneous detection of fluorescence in two different wavelength regions. In our experiments, we demonstrated a correlation between the lipophilicity of a camptothecin and its ability to accumulate in glioblastoma cells. Future studies will attempt to identify other important pharmacological correlates in tissues using multiphoton excitation. Since skin blood and tissues are translucent at long wavelengths, our results suggest the possibility of homogeneous or noninvasive clinical sensing of camptothecins with two-photon excitation (CA 63653).

Abstract ID: 517

Functional Molecular Imaging of Met Oncogene Product, A Novel Imaging Modality in Cancer

Galia Tsarfaty^1,2, Miriam Shaharabany³, Tammar Kushnir², Michal Firon³, Rotem Ron³, Yacov Itzchak², George F. Vande Woude¹, Ilan Tsarfaty^1,3

¹Van Andel Institute, Grand Rapids MI, ²Department of Diagnostic Imaging, MRI Institute, Sheba Medical Center, Tel-Hashomer, Israel, ³fBIT LTD, Rad Ramot Bio-Medical Incubator, Tel-Hashomer, Israel, and ⁴Department of Human Microbiology, Sackler School of Medicine, Tel Aviv University

The oncogene Met and its ligand HGF/SF play a significant role in the pathogenesis of breast cancers by increasing tumorigenicity and metastasis. The purpose of this work is to develop new modalities for functional molecular imaging (FMI) of oncogenes. DA3, a mammary adenocarcinoma cell line, was injected into the mammary gland of mice forming tumors expressing high levels of Met. HGF/SF or Heparin was intravenously injected and time-lapse gradient-echo images were acquired using 1.9-T MRI (Elscint). Met activation in vivo by HGF/SF alters the hemodynamics of normal and malignant Met-expressing tissues. Organs and tumors expressing high levels of Met showed the most substantial alteration in blood oxygenation levels as measured by blood oxygenation level dependent MRI (BOLD-MRI). Met-expressing tumors showed 30% alteration. No significant alteration was observed in tumors or organs that do not express Met. The extent of MRI signal alteration was also in correlation with HGF/SF doses. Doppler ultrasound measurements demonstrated that these MRI changes are at least partially due to altered blood flow. Heparin, which increases HGF/SF levels in the serum, has similar effects on normal organs and tumor hemodynamics. In autocrine models (Met-HGF/SF), we observed dramatic effects of Met activation in the tumor and normal organs. In inherited Met mutational activation model, we observed susceptibility of the normal organs and especially the tumor to Met signaling. This novel FMI modality could be used as a tool for the detection, analysis, and prognosis of a wide spectrum of tumors in which breast cancer is the first.

Abstract ID: 518

MicroPET Imaging of TAG72 Positive Human Xenografts Using Multistep Targeting

I. Guenther, P. M. Smith-Jones, R. Finn, B. Beattie, K. Campa, S. M. Larson

Nuclear Medicine Service, Department of Radiology, Memorial Sloan Kettering Cancer Service, New York, NY 10021

Intact monoclonal antibodies can specifically target tumors in vivo, but prolonged blood clearance of these agents limits their suitability as imaging agents with short-lived positron emitting radionuclides. In this study, we report on the use of multistep targeting to rapidly transport radionuclides to the tumor site using a carrier of DOTA–biotin. The DOTA–biotin conjugate was labeled with radiogallium using acetate buffer at 80–100°C. Radiolabeling yields were >99%, and >99% of this radioactivity could be specifically bound to avidin–agarose beads. Specific activities of 0.25–1 mCi/nmol were achieved. Mice and rats were injected with 5 million LS 174T tumor cells in Matrigel. After 7 days, the tumors were palpable and by 11 days large tumors were established. The animals were then infused with a streptavidin–CC49 conjugate. After 24 hr, a biotinylated N-acetylgalactosamine-containing “clearing agent” was administered to remove the circulating conjugate from the blood. Four hours later, either Ga-66, Ga-67, or Ga-68 labeled DOTA–biotin was administered to the animals. The animals were then imaged with a MicroPET or sacrificed for biodistribution studies up to 48 hr postinjection. Within 1 hr, the tumor/blood ratios were about 5 and the tumors were clearly visible in MicroPET images of mice and rats injected with Ga-66 or Ga-68. There was a slight increase in tumor accumulation, which was maximal at 3–4 hr postinjection (11.7 ± 4.4 %ID/g at 3 hr pi). At 3–4 hr postinjection, the only normal organs that could be visualized were the liver (0.56 ± 0.18 %ID/g), kidneys (1.72 ± 0.40 %ID/g), and bladder. The tumors could be visualized up to 24 hr postinjection, but there was a marked reduction in tumor uptake, which could be explained by this rapidly growing tumor model. This work demonstrates the ability of using short-lived generator-produced PET isotope to rapidly target tumor xenografts and the ability of a PET camera system to image high-energy positron emitters.

Abstract ID: 519

Dual Radiolabeling of N-[6-Fluoro-3-(3-Iodobenzoyl)benzyl]-2-Bromoacetamide with Fluorine-18 and Iodine Isotopes and its Conjugation with Oligonucleotides

Zizhong Li¹, Yushin Ding^1,2, Andrew Gifford¹, John Glass¹, Joanna Fowler², John Gatley¹

¹Medical Department and ²Chemistry Department, Brookhaven National Laboratory, Upton, NY 11786

Our goal is to evaluate the biological activity of oligonucleotides by various in vivo and ex vivo imaging modalities. We developed methods to introduce F-18 and iodine radioisotopes into a chemically identical organic molecule and used it to label oligonucleotides. N-[6-Fluoro-3-(3-iodobenzoyl)benzyl]-2-bromoacetamide was our target molecule. Its radioactive isotopo- logues: N-[6-[F-18]fluoro-3-(3-iodobenzoyl)benzyl]-2-bromoacetamide or N-[6-fluoro-3-(3-[I^*]iodobenzoyl)-benzyl]-2-bromoacetamide can be used to label oligonuleotides containing –SH group with either F-18 for PET imaging or I-125, I-123, and I-131 for high-resolution autoradiography, SPECT imaging, cancer therapy, and routine laboratory biodistribution studies. N-[6-Fluoro-3-(3-tributyltinbenzoyl)benzyl]-2-bromoacetamide was designed as the precursor for N-[6-fluoro-3-(3-[I^*]iodobenzoyl)benzyl]-2-bromoacetamide. It was synthesized in six steps starting from (3-tributyltin)benzylaldehyde. The radioiodine labeling was carried out by conventional electrophilic iodination. 4-(N,N,N-Trimethylamoniumtriflate)-3-cyano-3′-iodobenzophenone was the precursor for N-[6- [F-18] fluoro-3-(3-iodobenzoyl)benzyl]-2-bromoacetamide and was synthesized from 2-(N,N-dimethylamino)-5-bromobenzonitrile in four steps. The first step, nucleophilic F-18 labeling reaction, was carried out in acetonitrile at 80°C in 5 min and the radiolabeling yield was over 76%. Results from in vivo biodistribution study of N-[6-fluoro-3-(3-[I-131]iodobenzoyl)benzyl]-2-bromoacetamide in mice showed lack of significant accumulation of radioiodine in the thyroid, which was an indication of in vivo stability. Preliminary results showed a high yield of the conjugation reaction of N-[6-fluoro-3-(3-iodobenzoyl)benzyl]-2-bromoacetamide with phosphorothioate oligonucleotides. In conclusion, we have been successful in developing a dual-labeled organic group suitable for labeling phosphorothioate oligonucleotide.

Abstract ID: 524

The Sigma-1 Receptor PET Radiotracer [¹⁸F]FPS: Potential Advantages Over [¹⁸F]FDG for Tumor Visualization Demonstrated in Preclinical Models

Rikki N. Waterhouse¹, Rashid Fawwaz³, Lee Collier², Filip Dumont¹, Ronald Van Heertum³

Departments of ¹Psychiatry, ²Cardiology, and ³Radiology, Columbia University, New York, NY 10032

Recently, the expression of sigma-1 receptors has been reported to correlate significantly with Bcl2 expression and inversely with progesterone receptor expression in human breast cancers [S. Lafontaine et al. Br. J. Cancer, 2000, 82(12): p. 1958–66]. This provides evidence for the value of sigma-1 receptors as a prognostic marker in breast cancer. We have developed the selective sigma-1 receptor PET radiotracer, [¹⁸F]FPS, as an effective probe for labeling sigma-1 receptors in vivo. Data are presented to demonstrate that [¹⁸F]FPS holds promise as a tumor imaging agent, possessing several advantages over [¹⁸F]FDG especially in the breast and mediastinal area.

Methods: [¹⁸F]FPS (K_d = .8 nM) was synthesized as previously described (Collier et al. J Labelled Cmpds Radiopharm, 1996, 785–794). [¹⁸F]FPS and [¹⁸F]FDG were directly compared mice that spontaneously developed breast tumors. The effect of glucose levels on [¹⁸F]FPS biodistribution and tumor uptake was also examined in this model by administering dextrose intraperitoneally to one group of tumor-bearing mice. In addition, to determine if [¹⁸F]FPS will concentrate in tissues involved in inflammatory processes, [¹⁸F]FPS uptake in Lewis rats with rejecting heterotopically transplanted ACI rat hearts was examined.

Results: [¹⁸F]FPS uptake in breast tumors was high [10.7 ± 3.0 %ID/g (n = 5)] at 4 hr postinjection. The average tumor-to-tissue ratios at 4 hr for the blood, lung, muscle, heart, and liver were 39, 1.8, 9.1, 5.4, and 0.3 respectively. For [¹⁸F]FDG, peak tumor radioactivity concentration averaged 6 %ID/g and tumor/tissue ratios for the blood, lung, muscle, heart, and liver were 14, 2.5, 1.6, 0.7, and 10.1, respectively. In this model, [¹⁸F]FPS tumor and organ uptake was not affected by increasing glucose levels. In Lewis rats, [¹⁸F]FPS uptake in rejecting ACI hearts (0.30 ± 0.08 %ID/g, n = 5) was not significantly different from normal heart uptake (0.31 ± 0.05 %ID/g, n = 5). Histopathology confirmed the presence of a large degree of inflammatory components in all rejecting ACI hearts but not in normal Lewis hearts.

Conclusions: Due to the superior tumor/heart and tumor/muscle ratios obtained with [¹⁸F]FPS, this tracer is hypothesized to be superior to [¹⁸F]FDG for imaging tumors in the mediastinal area and potentially in the breast. In addition, the lack of change in [¹⁸F]FPS uptake in hearts affected by chronic inflammation and its unaltered distribution with increasing glucose levels indicates additional advantages of [¹⁸F]FPS over [¹⁸F]FDG.

Abstract ID: 525

[¹¹C]MCG: Synthesis and Uptake Selectivity of a Probe for NAALADase

Martin G. Pomper¹, John L. Musachio¹, Jiazhong Zhang², Paige Rauseo¹, Ursula Scheffel¹, John Hilton¹, Alan P. Kozikowski²

¹Department of Radiology, Johns Hopkins University School of Medicine, Baltimore, MD 21217 and ²Drug Discovery Program, Georgetown University Medical Center, Washington, DC 20057

Selective imaging of N-acetylated alpha-linked l-amino dipeptidase (NAALADase), an enzyme involved in regulation of excitatory signaling in the brain [1], may enable study of glutamatergic transmission presynaptically Furthermore, NAALADase has high sequence homology with the prostate-specific membrane antigen (PSMA), a Type II membrane glycoprotein that is primarily expressed in normal human prostate epithelium and is up-regulated in prostate cancer, including metastatic disease. We have synthesized [¹¹C] (S)-2-[3-((R)-1-carboxy-2-methylsulfanyl-ethyl)-ureido]-pentanedioic acid ([¹¹C]MeCys-C(O)-Glu or [¹¹C]MCG), an asymmetric urea that is a potent (K_i = 1.9 nM) inhibitor of NAALADase, as a potential imaging agent for NAALADase/PSMA. Radiosynthesis (NCA) was performed by C-11 methylation of the free thiol in DMF/NH₃ (45°C/3 min). [¹¹C]MCG was obtained in high specific radioactivity (>148 GBq/umol) and acceptable radiochemical yield (12% nondecay) within 30 min after e.o.b. At 30 min postinjection, [¹¹C]MCG showed 32.99 ± 5.14, 0.45 ± 0.13, and 1.09 ± 0.22 %ID/g in mouse kidney, which displays high levels of NAALADase/PSMA, muscle and blood, respectively. Little radioactivity (0.12 ± 0.03, 0.06 ± 0.0,0.08 ± 0.02 %ID/g for brainstem, cortex, and cerebellum, respectively) gained access to the brain. Blockade with either unlabeled MCG (5–1000 μg/kg) or PMPA, another potent inhibitor of NAALADase [2], gave a smooth dose–saturation curve with a 5-fold reduction in binding in target tissue (kidney) at 100 μg/kg and near absence of binding at the highest dose of blocker. [¹¹C]MCG is highly stable in the mouse, showing 3.5% and 2.0% metabolism in plasma and kidney, respectively, at 30 min. These results suggest that [¹¹C]MCG may provide a suitable imaging agent for NAALADase/PSMA in the periphery. [Supported by CA92871 to M.G.P.]

Abstract ID: 526

Development of Technetium-99m and Indium-111 labeled p185HER2/neu Specific Peptidomimetic Radiopharmaceuticals for Tumor Imaging and Therapy

Zifen Su^1,2, Alan Berezov¹, Qindu Liu¹, Abas Alavi², Mark I. Greene¹, Ramanchandran Murali¹

¹Department of Pathology and Laboratory Medicine, University of Pennsylvania, 3620 Hamilton Walk and ²Division of Nuclear Medicine, The Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104

The p185HER2/neu oncoprotein, a tyrosine kinase receptor, is overexpressed in ~35% of human breast, ovarian, prostate and a higher percentage of pancreatic cancers. Anti-HER2 monoclonal antibody, Hercept (Genetech), is currently used in the advanced breast cancer therapy. A small molecule AHNP has been shown to mimic the monoclonal antibody (Park et al., Nat Biotechnol 18, 194–198, 2000.) with comparable affinity and specificity. AHNP is expected to be superior over antibody for radioimaging due to its smaller molecular weight and more rapid clearance rate. Accordingly, ^99mTc-labeled AHNPs both as tumor imaging and ¹¹¹In labels as imaging and therapeutic agents are being developed in this laboratory. Two second-generation forms of AHNPs have been designed in this laboratory and are suitable for radiolabeling. We have characterized the p185HER2/neu binding specificity and affinity (determined by Biacore analysis) in the range of 100 nM. DTPA as a chelator of ^99mTc and ¹¹¹In has been successfully conjugated to the new peptidomimetic species. One or two DTPA molecules can be conjugated to the peptidomimetic. The molecular weights of the purified conjugates were then confirmed by MALDI-TOF mass spectrometry. Indium complexes of the DTPA-AlaAHNP and DTPA-AbuAHNP have also been prepared, purified, and characterized by mass spectrometry. The results of radiolabeling T6-17, a cell transformed by HER2/neu, and T6-17 xenografted animal imaging tests of the radiolabeled AHNPs will be presented.

Abstract ID: 527

Characterization of the Tumor Suppressor Protein, RASSF1A

Shairaz Baksh, Julian Downward

Department of Signal Transduction, Imperial Cancer Research Fund, 44 Lincoln Inn's Fields, London WC2A 3PX

Small cell lung cancer (SCLC) patients have specific alterations to chromosome 3. Genetic analysis of the short arm of chromosome 3 revealed that deletion of several lengths of nucleic acid sequence exists. RASSF1A was found to be selectively deleted and the promoter silenced by methylation in 70–80% of SCLC patients, while isoforms B and C remained unaffected. RASSF1A was also found to be deleted in 40% of breast carcinomas, 70% of renal carcinomas, 40% of non-SCLC, and 10% of ovarian carcinomas. We investigated the function of RASSF1A in two human transformed cell lines—U2OS (a human osteosarcoma cell line) and MCF-7 (a breast cancer cell line). Functional analysis revealed that RASSF1A can homodimerize with itself and is predominantly localized to “microfilament”-like structures. The amino terminal of RASSF1A was as potent as the full-length protein in promoting CD95- or staurosporineinduced apoptosis in MCF-7 cells. RASSF1A has a consensus 14-3-3 binding site and was found to selectively interact with 14-3-3 eta and zeta. Furthermore, this interaction may function to localized RASSF1A to the cytoskeletal network surrounding the nuclear membrane and function to control gene transcription of at least three p53-responsive targets—cyclin A, p21, and possibly cyclin G1. We postulate that the RASSF1A is an important tumor suppressor element that may interact with the cytoskeletal network and regulate apoptosis of cells that may have an abnormal cytoskeletal environment.

Abstract ID: 534

Molecular Imaging Using Sodium/Iodide Symporter (NIS) Gene in Nude Mouse Model of Human Colon Cancer Cells

J. H. Shin, Y. J. Lee, J.-K. Chung, J. J. Min, J. M. Jeong, D. S. Lee, M. C. Lee

Department of Nuclear Medicine Seoul National University College of Medicine, Seoul, Korea

Radioiodine scanning, which has been performed conventionally for thyroid disease, depends on the sodium iodide symporter (NIS). To investigate the feasibility of NIS as an imaging system, we transfected rat NIS (rNIS) genes into a human colon cancer cell line (SNU-C5) by lipofection. NIS gene transfection is confirmed by RT-PCR. The uptake and efflux of ¹²⁵I were estimated in the transfected cell line (SNU-C5N). To assess the efficiency of rNIS expression in vivo, we injected SNU-C5 and SNU-C5N cells in the right and left flank of nude mice, respectively. Biodistribution study and tumor imaging with ¹³¹I scintigraphy were carried out. SNU-C5N cells resulted in ¹²⁵I uptake by these cells to a level 10 times higher than control cells (SNU-C5). For first 10 min, up to 70% of the cellular radioactivity was released into the medium. In the biodistribution study using SNU-C5N-xenografted mice, the %ID/g of SNU-C5N tumors at 1, 3, 6, and 12 hr after injection of ¹²⁵I were 4.43%, 1.09%, 1.05%, and 0.05%, respectively, which were significantly higher percentages than those for SNU-C5 tumors (p<.05). In tumor imaging, SNU-C5N-xenografted tumor was clearly visible. Our study shows that radioiodine imaging with NIS gene transfection can be used for tumor imaging. After further study, this system can be applied for monitoring of gene therapy and other gene transfer.

Abstract ID: 535

PET-CT Alters the Management of Patients with Gynecologic Malignancies

D. E. Heron¹, A. Sit³, G. King¹, J. Yap², K. Gerszten¹, R. Selvaraj¹, J. Comerci³, R. P. Edwards³, C. C. Meltzer², T. Blodgett², D. W. Townsend², H. Gallion³

Departments of ¹Radiation Oncology, ²Radiology, and ³Division of Gynecology Oncology, University of Pittsburgh School of Medicine, UPMC Magee-Womens Hospital, Pittsburgh, PA

Purpose: Positron emission tomography (PET) is an emerging imaging technology for the diagnosis and staging of cancer with significant implications in clinical oncology. However, PET lacks the anatomical detail customary to computed tomography (CT) which is required for radiation treatment planning. To evaluate the impact of fluorodeoxyglucose 18FDG (18FDG) PET-CT on the management of patients with primary and recurrent gynecology malignancies, we performed radiation therapy planning using a planning abdominopelvic CT (pCT) versus PET-CT prospectively.

Materials and Methods: Ten patients with primary and recurrent gynecologic malignancies were evaluated for radiation treatments at the University of Pittsburgh School of Medicine, Magee-Womens Hospital, Pittsburgh, PA. Seven patients with cervical cancer (five primary, two recurrent), two with endometrial cancer (one primary, one recurrent), and one patient with ovarian cancer were evaluated with pCT. Prior to obtaining a PET-CT to evaluate for occult nodal and soft tissue metastases, all pCT studies were independently reviewed by a gynecologic radiologist and were used to define the CT-based treatment plan. PET-CT images were acquired on an integrated PET-CT scanner (CTI PET Systems) at 1 hr following intravenous injection of approximately 6–10 mCi 18FDG and were reviewed by a radiologist/nuclear medicine physician. All patients were treated with either conventional 3-D radiotherapy (3-D CRT) or intensity-modulated radiotherapy (IMRT) on a high-energy (>10 MV photons) linear accelerator.

Results: All patients had uptake in the primary or index site of known disease on PET-CT. Of the 10 patients with primary or recurrent disease, four had significant modification made to their planned radiation portals. Using IMRT, one patient had expansion of her para-aortic portals to encompass the extent of increased FDG uptake; two patients required the addition of a para-aortic field as a result of findings on PET-CT. One of these patients also required an additional pelvic field to treat an occult recurrence involving the left pelvic side-wall at a previously positive biopsy site. Two patients had a simultaneous boost of abnormal pelvic nodes with IMRT based on the PET-CT findings.

Conclusions: PET-CT is a rapidly emergent functional imaging technology. In our group of patients with primary and recurrent gynecologic malignancies, nearly half had significant modification of their planned radiation therapy fields based on findings on PET-CT. Further investigation is necessary to determine the sensitivity and specificity of 18FDG PET-CT in gynecologic malignancies and its impact on radiation treatment planning.

Abstract ID: 562

Bioanatomic 3-D Radiation Treatment Planning for Brain Tumors: Concept and Study

J. D. Bourland, E. G. Shaw, L. P. Adler, B. A. Harkness, J. H. Burdette

Departments of Radiation Oncology and Diagnostic Radiology, Wake Forest University, Winston-Salem, North Carolina, USA

Conventional 3-D radiation treatment planning for brain tumors uses anatomical information from CT and MR images to identify the tumor and surrounding tissues for radiation treatment planning. A conventional plan displays the patient's anatomy and the calculated physical dose distribution. However, tumor biology is not modeled or represented, even though tumor radiation response depends on tumor biology and not just on anatomy and the radiation time–dose schedule. Two important characteristics of tumor biology include tumor oxygenation and tumor blood flow. In a pilot study using positron emission tomography (PET) imaging, we used F-18 Misonidazole (hypoxia) PET imaging and O-15 Water (perfusion) PET imaging to measure the spatial distributions of oxygen and blood flow in human brain tumors (biopsy-proven glioblastoma multiforme, or GBM). In this study, the biological images of tumor hypoxia and blood flow from PET are combined with anatomical images from CT and MRI to comprise a biological–anatomic representation of the patient's tumor, providing an opportunity for biologically directed treatment optimization. A key aspect includes the use of intensity-modulated radiation treatment (IMRT) to deliver nonhomogenous but biologically significant doses to the target volume. We have given this approach and concept the name “Bioanatomic” and are investigating its value for optimizing treatment for patients with malignant brain tumors. The Bioanatomic Radiation Treatment Planning concept, PET hypoxia and perfusion imaging protocols, and CT/MR/PET imaging results for four GBM patients in a pilot imaging study are presented. [Supported in part by North Carolina Baptist Hospital Technology Grant A-01-97 and by a grant from Varian Medical Systems.]

Abstract ID: 579

A Novel Trispecific Antisense Oligonucleotide to Inhibitors of Apoptosis Proteins Enhances Chemosensitivity in the T24/83 Bladder Cancer Cell Line

Sam Gray¹, Brian Duggan², Uwe Zangemeister-Wittke³, Kevin McEleny⁴, Neil Anderson⁵, Patrick Keane², Robin Johnston², Kate Williamson¹

¹Queen's University Belfast, ²Belfast City Hospital, ³UniversitätsSpital Zürich, ⁴University College Dublin, and ⁵Royal Group of Hospitals Trust, Northern Ireland

Introduction: Our group has reported enhanced synergistic chemosensitivity to mitomycin c after pretreatment with either G3139 or 2009 antisense oligonucleotides to Bcl-2 in the T24/83 bladder cancer cell line. Inhibitors of apoptosis proteins (IAPs) are a group of structurally homologous proteins that function to inhibit the caspases and prevent apoptosis. The aim of this study was to examine the effects of a novel antisense oligonucleotide designed to be trispecific for c-IAP1, c-IAP2 and XIAP on the T24/83 cell line. Cells were incubated with the antisense oligonucleotides and mitomycin c for 1 hr to mimic the clinical application.

Methods: Twenty thousand T24/83 cells were seeded onto 120 coverslips. Cells were grown for 48 hr under standard tissue culture conditions in 12-well plates. To groups of six wells, 500 ml antisense oligonucleotide to IAP (100 nM), antisense oligonucleotide to IAP (200 nM), scrambled antisense oligonucleotide (100 nM), or scrambled antisense oligonucleotide (200 nM) were added. To groups of 12 wells, lipofectin (2000 nM), lipofectin (4000 nM), antisense oligonucleotide to IAP (100 nM) plus lipofectin (2000 nM), antisense oligonucleotide to IAP (200 nM) plus lipofectin (4000 nM), scrambled antisense oligonucleotide (100 nM) plus lipofectin (2000 nM), or scrambled antisense oligonucleotide (200 nM) plus lipofectin (4000 nM) were added. Twenty-four wells had no treatment. Medium in all wells was replaced after 1 hr. After a further 23 hr, mitomycin c (200 μg/ml) was added to 48 wells for 1 hr. Coverslips were Giemsa stained after either a further 3 or 24 hr. An apoptotic score was assessed independently by two observers from triplicate coverslips for each time point and concentration within each of the 10 treatment groups. Scores for apoptosis, apoptotic dust, and cell density were summed. Apoptosis ranged from 1 = no evidence to 5 = >10 apoptotic cells/high power field. Apoptotic dust ranged from 1 = no evidence to 4 = frequent areas in each high power field. Cell density ranged from 1 = no treatment density to 3 = significant decrease compared to no treatment. Apoptotic scores were statistically compared using Kruskal–Wallis and Mann–Whitney tests.

Results: Apoptotic scores ranged from 3 to 12. Cells treated with antisense oligonucleotides with or without lipofectin had apoptotic scores <6 at both time points. There was no significant difference between cells treated with mitomycin c only and those treated with scrambled antisense oligonucleotide plus mitomycin c after 24 hr (apoptotic scores = 9 and 9.4, respectively). However, after 24 hr, there was a significant difference between cells treated with 100 nM antisense oligonucleotide to IAP plus mitomycin c (apoptotic scores = 11.3 and 11.0) and those treated with mitomycin c alone (p = .043 and p = .034, respectively, for the two observers). There was a also a trend for the cells treated with scrambled antisense oligonucleotide plus mitomycin c to differ from those treated with antisense oligonucleotide to IAP plus mitomycin c.

Conclusions: Antisense oligonucleotide to IAP enhanced chemosensitivity to mitomycin c in the T24/83 cell line.