Abstract
EXTENDED ABSTRACT Intravasation and Metastatic Seeding - James P. Quigley, The Scripps Research Institute In Vivo model systems for cancer metastasis that are utilized for discovery-based approaches to screen for and identify critical molecules or rate limiting processes should fulfill a number of criteria: (1) They should recapitulate, fairly accurately, the multi-step metastatic cascade; (2) for screening purposes, the models should be relatively rapid (days or weeks not months or years), (3) they should be quantitative, especially if rate limiting reactions or molecules are to be determined, and (4) the models should be experimentally facile so that molecular or observational or therapeutic intervention during the entire metastatic cascade is feasible. The chick embryo model for tumor dissemination clearly fulfills these criteria and is an ideal system for discovery-based approaches. The highly-vascularized chick embryo\#8217;s chorioallantoic membrane (CAM) efficiently supports the growth of inoculated xenogenic tumor cells as they go on to develop 100-500 mg primary tumors and actively disseminate to nearby and distant tissues, all in a 5-8 day period. Thus the model is the most rapid, in vivo metastasis model known. The dissemination of human tumor cells in the chick embryo can indeed be shown to undergo all the distinct steps in the metastatic cascade including invasion and escape from the primary CAM tumor, intravasation into the vasculature, survival and arrest in the circulation, extravasation into secondary tissue and finally formation of micrometastatic foci. The actual number of disseminated human tumor cells in the tissues, and the rate of their appearance in these tissues can be quantified by qPCR methods employing primate-specific alu repeat sequences, which are completely absent in avian DNA thus yielding an extremely low background. Therefore, the data generated with this model is reproducible, quantifiable and readily useable for identifying rate limiting reactions. Furthermore, the dense capillary network of the CAM also serves as an initial repository for aggressive tumor cells that escape from the primary tumor and intravasate into the host vasculature. This spontaneous metastasis setting thus provides a unique model to study in vivo, specifically the intravasation step of the metastatic cascade. In addition, during experimental metastasis when tumor cells are inoculated intravenously, the CAM capillary system serves as a site for initial arrest and subsequently for tumor cell extravasation and colonization, the final steps in the metastatic cascade. Thus, the molecular and tissue composition of the CAM and its accessibility for experimental intervention makes chick embryo CAM systems attractive models to follow the fate and visualize microscopically the behavior of various types of grafted tumor cells in both spontaneous and experimental settings. Our laboratory has utilized this model to molecularly and microscopically analyze specific steps in the cascade. Two distinct experimental approaches have been taken: One approach involves the in vivo selection of human tumor cell congenic variants which differ widely in their ability to carry out a specific step in metastasis. A second approach utilizes functional proteomics coupled with subtractive immunization to generate and select unique function-blocking mAbs that inhibit individual steps in metastasis and allow for identification of the functionally relevant antigens. This presentation will highlight these two approaches and will focus on intravasation, probably the least-studied step in metastasis. Two human tumor cell variants (HT-hi/diss and HT-lo/diss) were selected from the parental HT 1080 fibrosarcoma cell line by a series of in vivo and in vitropassages. The HT-hi/diss variant exhibited a 50-100 fold enhanced intravasation rate over that of HT-lo/diss and manifested a corresponding enhancement in spontaneous metastasis. Importantly, the two variants were almost identical in their CAM and organ (lung, liver, brain) colonization rate during experimental metastasis. These results indicate that the variants differ mainly in their capabilities of carrying out the early events in the metastatic cascade. These distinctive characteristics of the two variants were confirmed in SCID mouse models for spontaneous and experimental metastasis indicating that the resulting selected properties were not unique to the chick CAM model. Direct comparisons of the variants\#8217; in vivo cellular behavior patterns at the morphological level indicated that the HT-hi/diss variant escaped from the primary CAM tumor at greatly enhanced levels over that of HT-lo/diss cells migrating out from similar sized primary tumors. Most interestingly, the escaping HT-hi/diss cells exhibited a form of \#8220;vasculotropism\#8221; in which the tumor cells appeared to migrate towards and/or wrap themselves around blood vessels. This was in contrast to the lower number of escaping HT-lo/diss cells which were distributed randomly in the CAM mesoderm. The use of fluorescent-tagged tumor cells (green), and differently tagged vasculature (red), along with live cell imaging techniques allowed us to affirm and quantify the vasculotropism manifested by escaping HT-hi/diss tumor cells. The data strongly suggest that it is a combination of the invasive and vasculotropic behavior of the HT-hi/diss cells that lead to their high intravasating ability. In addition to expressing proteins that drive the HT-hi/diss tumor cells toward, onto and into blood vessels, these tumor cell variants also are able to drive blood vessels toward them, in that HT-hi/diss cells are substantially more angiogenic than their congenic non-malignant counterparts, HT-lo/diss cells. Again, microscopic imaging and also quantitative scoring of new blood vessel formation allowed us to confirm the high angiogenic-inducing ability of the HT-hi/diss variant. The striking differences in the cellular/morphological behavior of the selected tumor variants, along with the differential in their spontaneous disseminating ability, allows for probing differences in specific protein expression patterns that might correspond directly to their tumor behavior properties. We have employed various functional proteomic approaches in tandem with our live cell imaging analyses in order to identify distinct proteins that contribute to a specific step or process in metastatic dissemination. One such approach known as Activity Based Proteomic Profiling (ABPP), which utilizes fluorescent-tagged chemical probes that only react with activated proteolytic enzymes and not their widely-expressed inactive zymogens. Utilizing this profiling approach we determined that both of our HT 1080 fibrosarcoma variants express equal levels of the zymogen form of urokinase type plasminogen activator (pro-uPA), but the HT-hi/diss variant differentially activates single-chain pro-uPA to two-chain active uPA. Inhibition of active uPA with specific blocking reagents directly added to HT-hi/diss CAM tumors leads to a significant inhibition of the intravasation step, implicating active uPA as a functional contributor to early steps in the metastatic cascade. Another functional proteomic approach that we have utilized is subtractive immunization. In this whole cell immunization approach, mice are first immuno-tolerized to the non-metastasizing HT-lo/diss cell variant, followed by immunization with the metastasizing HT-hi/diss variant. This subtractive approach is used in order to generate unique monoclonal antibodies (mAbs) directed against cell surface antigens that are distinctly enhanced on the HT-hi/diss variant. The initially selected mAbs are then screened in an unbiased manner for their function-blocking capabilities i.e. their ability to inhibit HT-hi/diss dissemination in the chick embryo model. Identification by immuno-proteomic approaches of the antigen recognized by a specific, metastasis-inhibitory mAb will implicate the identified antigen as functionally involved in metastasis. One of our recently identified antigens, the tetraspanin CD151, has now been shown to function in an early step in metastasis, namely escape from the primary tumor. Another antigen identified by this immuno-proteomic approach is CUB domain containing protein-1 (CDCP1), which appears to function as a pro-survival molecule after intravasating tumor cells arrest in the vasculature. In conclusion, chick embryo CAM models have proven to be invaluable tools and experimental systems to study individual steps of human tumor cell dissemination. The major advantages of these models include quantitation, rapidity, the accessibility for repetitive intervention and the unique ability for intravital imaging and efficient molecular, kinetic and morphological analysis of tumor cell dynamics. 1
Citation Information: In: Proc Am Assoc Cancer Res; 2009 Apr 18-22; Denver, CO. Philadelphia (PA): AACR; 2009. Abstract nr SY17-3.
100th AACR Annual Meeting-- Apr 18-22, 2009; Denver, CO