New strategies for discovery of accessible tumor targets

It is clear that reducing tissue complexity to a manageable subpopulation ofthe cell types of interest is necessary. Already, much work has been done to develop computer software and statistical approaches to analyze and reduce the complexity of the data generated by global genomic and proteomic experiments. Another approach that might be more successful is, rather than attempt to slog through the mountains of data produced at the end of typical experiments, to reduce tissue sample complexity at the beginning, before high-throughput analyses are applied. This will require new analytical paradigms to focus the power of global, high-throughput identification technologies.

One widely used approach to reducing sample complexity has been to study tumor cells that have been isolated and grown in culture. Unfortunately, both enzymatic/mechanical tissue disassembly and growth in culture contribute to phenotypic changes that alter native cellular function and protein expression (36). Direct comparisons of protein expression have revealed <25% similarity between primary tumor cell biopsies vs cultured, clonally selected prostate tumor cell lines vs patient-matched microdissected cell-populations-derived biopsy specimens (37). For example, in vitro caveolin-1 expression is downregulated in transformed cell lines and in various cultured human cancer cell lines, and induced expression greatly retards tumor cell growth (38-44). Yet, in vivo, caveolin-1 can be highly expressed in tumors and is associated with increased tumor cell survival, aggressiveness, metastatic potential, suppression of apoptosis, acquisition of multidrug resistance, and resistance to chemotherapy (42,45-52). Thus, using cultured tumor cell lines exclusively for discovering tumor markers may cause potentially useful markers to remain undetected.

In response to these difficulties, new technologies, such as laser capture microdissection, have been developed and optimized to isolate specific cell types. By permitting the selection of specific cells from a tissue, laser capture microdissection has the advantage of allowing researchers to focus directly on the cell subpopulation of interest in the absence of contaminating cells so that different cell types can be compared directly. However, the complexity within a single cell is still considerable enough to confound attempts at target discovery. Moreover, even with the discovery of cell-type-specific targets, those targets might not be suitable for selectively targeting a single organ or diseased tissue such as solid tumors in vivo. Most tissue- and disease-associated gene products are expressed by the cells primarily constituting the tissue (e.g., tumor cells) and, as such, these potential targets remain inside tissue compartments that usually are not readily accessible to biological agents, such as antibodies injected into the circulation. This inaccessibility prevents the development of site-directed therapies (5-7,53,54) as well as hinders many molecular imaging agents from revealing important structural and functional information in vivo (1-4). A notable exception has been hematological tumors, such as lymphomas and leukemias, where antibodies can access and, in fact, target the tumor cells in vivo, thereby rendering such immunotherapies quite effective (55,56). However, in solid tumors, multiple barriers to delivery prevent similarly effective treatment in vivo, although many immunotherapies work quite effectively and specifically in vitro (10,53-57).

Because of these accessibility issues, one meaningful subset of the tumor tissue that might inherently be worth pursuing comprises the cells constituting the critical tissue-blood interface immediately accessible to agents absorbed or injected into the circulating blood. Endothelial cells line the blood vessels of all tissues to form an interface as an attenuated cell monolayer that plays a significant role in controlling the passage of blood molecules and cells into the tissue and in many physiological functions, including vasoregulation, coagulation, and imflammation, as well as tissue nutrition, growth, survival, repair, and overall organ homeostasis and function (58). Disruption ofthe vascular endothelium and its normal barrier function can lead rapidly to tissue edema, hypoxia, pathology, and even organ death (59). Thus, the endothelium forms one in vivo barrier that prevents many current imaging agents, drugs, and gene vectors from optimal access to target cells within organs (10,60-63).

The endothelia of blood vessels are heterogeneous and might express potential tissue and/or disease-specific targets. Yet most ofthe endothelial cell is again not inherently and fully accessible to agents circulating in the blood. Only the luminal endothelial cell surface is directly in contact with the circulating blood and thus appears to be ideally suited for tissue- and tumor-specific targeting. Yet, discovering useful targets on this surface has not been readily achieved, primarily for technical reasons. The endothelium constitutes a very minor component ofthe tissue that makes it difficult to study in vivo using genomic or proteomic analytical techniques because the relatively small amount of a potentially useful target protein associated with the endothelial cell surface might be obscured by "molecular noise" from highly abundant, unrelated material originating from other much more abundant cell types within the tissue and, thereby, escape detection.

Multiple advantages for targeting tumor endothelium relative to tumor cells are evident in theory. All tissues in the body and especially tumors rely on the bloodstream to supply critical nutrition. The endothelial cell surface is freely and immediately accessible through the blood circulation, whereas tissue and tumor cells reside inside the tissue on the other side ofthe vascular wall and are, for the most part, inaccessible (10,54,64). The endo-thelium elaborates specific mechanisms for selecting molecules to overcome its natural barrier function, which can be usurped to deliver drugs and gene vectors inside the tumor tissue. For example, caveolae are specialized invaginations abundant at the surface ofthe vascular endothelium that can mediate transport into the cell (endocytosis) and transport across the cell monolayer (transcytosis) (11). Antibodies that target tissue-specific endothelial cell proteins in caveolae can rapidly penetrate the endothelial cell barrier to reach underlying tissue cells (65). In addition, endothelial cells have a stable genome, so that, unlike tumor cells, they are unlikely to undergo substantial mutation to permit the outgrowth of drug- and targeting-resistant cells. Finally, because each microvessel provides oxygen and nutrients for large numbers of underlying tissue cells, there is an inherent amplification mechanism that comes even from the destruction of one endothelial cell, which naturally initiates local coagulation to occlude vessels serving the hundreds and even thousands of tumor cells.

There is growing evidence that tumor-induced molecular targets might exist on the endothelial cell surface of blood vessels feeding tumors. The microenvironment ofthe tissue surrounding the blood vessels can greatly influence the phenotype of the endothelial cells (36,66-72). Although the degree to which normal and neoplastic tissues can modulate endothelial cell surface protein expression in vivo is unknown, indirect evidence of molecular heterogeneity of endothelial cells in different organs and even solid tumors comes from the reported ability of select cells and even bacteriophage displaying select peptide sequences to home to specific tissues after intravenous injection (73,74). Genomic analysis of human endothelial cells isolated from enzymatically digested neoplastic tissue has provided several angiogenic markers of unknown subcellular localization (75).

Neither global protein nor genomic analysis of whole tissue is likely to discover endothelial cell targets in large part because the endothelial cells are such minor components of the tissue. Subcellular fractionation of the tissue is critical to unveil the protein concentrated in the endothelial cell surface caveolae that otherwise are beyond the dynamic range of detection, effectively "swamped out" by the much stronger signals from nearly 100,000 other, more abundant, proteins. Moreover, endothelial cells become more difficult to analyze because of their exquisite sensitivity to their natural environment in tissue. Endothelial cell responsiveness to a changing tissue microenvironment (e.g., during tumor-igenesis) is actually the underlying basis of their tissue-specific qualities and the targets being sought. Past approaches that attempted to define the molecular topography of the vascular endothelium have relied primarily on analyzing endothelial cells isolated from the tissue through enzymatic digestions that disassemble the tissue sufficiently to release single cells that are then usually sorted using endothelial cell markers (75-77). The study of isolated endothelial cells has yielded much functional and molecular information; however, both the significant perturbation of the tissue and the growth in culture contribute to morphologically obvious phenotypic drift that can translate rapidly into loss of native function and protein expression (36,66).

Lastly, another important stumbling block to effectively identifying potential intravenously accessible target molecules at the critical blood-tissue interface is the methodological difficulty in examining endothelium in vivo. Even in highly vascularized tissue, the endothelium represents only a very small percentage ofthe cells constituting the tissue, and the proteins of the endothelial cell membrane exposed directly to the blood represent only a small fraction of the total proteins of this cell type and even more so of the total tissue. Classic techniques for the isolation of plasma membranes from tissues will yield a membrane fraction that will contain endothelial cell membranes, but only as a small percentage ofthe total membrane isolated. This makes detection and identification of endothelial-spe-cific proteins very difficult and comparisons between normal and neoplastic tissues nearly impossible.

We have developed a new approach that facilitates the isolation ofthe luminal cell surface as it exists in its native state in the tissue in vivo by purifying the luminal endothelial cell plasma membranes directly from tissue (78). We perfuse the tissue microvasculature in situ with a positively charged, colloidal silica solution that selectively coats the luminal endothelial cell membrane normally exposed to the circulating blood and creates a stable adherent silica pellicle marking this specific membrane of interest (66,79-81). Such a coating increases the plasma membrane's density and is so strongly attached to the plasma membrane that after tissue homogenization, large sheets of silica-coated endothelial cell plasma membrane can be readily isolated away from other cellular membranes and debris by centrifugation through a high-density medium. This technique provides endothelial cell plasma membranes isolated with little contamination from intracellular components or even the plasma membranes of other lung tissue cells (66,79-82). As such, it is a reasonably ideal starting material for proteomic analysis,

0 0

Post a comment