Abstract
To facilitate detection of tumor cells at the highest resolution in any organ in athymic nude mouse model systems, a histochemical marker gene [bacterial IacZ or human placental alkaline phosphatase (ALP)] was transfected into specified transformed/tumor cells (fibrosarcoma or neuroblastoma). The fates of tumor cells were followed qualitatively and quantitatively by histochemical staining of whole organs or organ sections. Primary tumors developed initially via formation of “curly-haired” complexes of cells in the subcutis or dermis, followed by division of a large fraction of cells. When two tumor classes were mixed before injection, outgrowth occurred in regional concentrations of the primary tumor. Blood microvessels were detectable within 72 hr of injection, growing into tumor regions. IV injection routinely yielded multicellular foci in the lungs within minutes as precursors of experimental metastases. Micrometastasis was further resolved with cells “inactivated” by different treatments and by co-injection of two different tagged cell types. These approaches using different histochemical marker genes to “tag” different tumor cell classes, along with more advanced molecular biological approaches, permit us to characterize gene expression and its reversibility during the earliest stages of primary tumor formation and micrometastasis to virtually any organ in the recipient animal.
Keywords
B
Our laboratory had been investigating the role(s) of the ras oncogene in promoting metastasis of transfected Balb/c 3T3 cells, particularly to the lung, in athymic nude mice (Culp et al. 1991). In parallel, studies were initiated on the role of the amplification of the N-myc oncogene in human neuroblastoma in altering the extracellular matrix adhesion properties during metastatic spread to bone, bone marrow, and liver (Culp and Barletta 1990). In both systems we were frustrated by our inability to detect single tumor cells or small collections of tumor cells at ectopic or orthotopic injection sites, as well as at multiple target organs for metastasis. Detection of single tumor cells would permit more effective analyses of quantitative relationships in tumor progression, more effective resolution of the relationships between the tumor cells and “normal” host tissue cells in their environment, and two-way communication between tumor cells and normal tissue cells that ultimately would determine whether tumor cells die, stabilize but do not divide over long periods of time, or begin to divide rapidly.
During this period of frustration, Sanes and collaborators (1986) published the first use of the bacterial IacZ gene as a histochemical tag to track the lineage of specific cells in the developing embryo. This important study piqued our interest in the possibility of using this gene as a genetic and histochemical “tag” to track tumor cells in our animal model studies. The first studies for using this gene in any tumor system were reported by our laboratory, in collaboration with the laboratories of Thomas and Theresa Pretlow at this University with their expertise for pathological analyses, for quantitatively and qualitatively evaluating the metastatic spread of ras-transfected 3T3 cells (Lin et al. 1990a,b). There was also considerable rationale for using histochemical marker genes rather than fluorescing gene products such as luciferase or, more recently, green-fluorescent protein. Histochemical staining permits detection of the tumor cells by light microscopy and better evaluation of the topological relationship of the tumor cells to neighboring normal tissue cells and to blood vessels. With fluorescent tumor cells, these relationships are difficult to resolve. Histochemical gene products were also remarkably tough enzymes, resistant to proteolytic breakdown in tumor cells. Their colored products were distinctive, easily identified, and resistant to degradation with proper storage of samples. Furthermore, their histochemical products were retained stably within the confines of the tumor cell and did not readily dissipate into the environment. These principles applied to our identification of two other (alternative) histochemical marker genes, human placental alkaline phosphatase (ALP) gene and the Drosophila alcohol dehydrogenase gene (Lin and Culp 1991). Of particular note is that this ALP gene was membrane-localized, permitting enzymatic activity to be retained within “tagged” tumor cells. To avoid further genetic instability caused by retroviral insertion into tumor cells that are already genetically unstable, these genes were inserted into integrating plasmids, with expression being driven by the RSV LTR or the cytomegalovirus promoters (Lin and Culp 1991).
This article outlines various uses of these histochemical marker genes in both mouse fibrosarcoma and human neuroblastoma model systems. The fibrosarcoma system includes tumorigenic/nonmetastatic sis oncogene-transfected Balb/c 3T3 cells and highly metastatic ras-transfected 3T3 cells. Human Platt neuroblastoma cells were analyzed in parallel studies to define cell type and tumor type specificity.
Materials and Methods
Cells and Growth Conditions
The LZEJ and APSI cell lines have been described (Lin et al. 1990a,1992). LZEJ is a clone of H-ras-transformed Balb/c 3T3 cells that were subsequently transfected with an integrating plasmid bearing the RSV LTR-regulated IacZ gene. APSI is a clone of human sis-transformed 3T3 cells subsequently transfected with an integrating plasmid harboring the RSV LTR-regulated ALP gene. These lines stably express their respective marker genes for 10–15 passages but eventually yield a higher proportion of the population that are nonexpressing. All animal studies used early-passage cells that were grown as described (Lin et al. 1990a,1992). LacZ-transfected (RSV LTR-regulated) human Platt neuroblastoma cells (LZPt clone 1, 2, or 3) were also described previously (Kleinman et al. 1994).
Animal Tumorigenesis Studies
Athymic nude mouse studies were approved by the Animal Care and Use Committee of Case Western Reserve University. They were conducted in the AAA-LAC-acredited Athymic Animal Facility of the Cancer Center of the School of Medicine. LZEJ or APSI cells were injected sc (a pseudo-orthotopic route for fibrosarcoma cells) or IV via the tail vein (ectopic route) to analyze experimental metastasis. Whereas LZEJ cells spontaneously metastasize to lungs from the sc site (Culp et al. 1991), APSI cells do not metastasize to any organ. LZPt cells were injected either sc or intradermally (ectopic sites) or via the adrenal gland (an orthotopic route for neuroblastoma) (Flickinger et al. 1994). Details of tumor studies have been described previously (Lin et al. 1990a,b, 1992; Lin and Culp 1992a; Kleinman et al. 1994; O'Connor and Culp 1994).
Histochemical Staining of Tissues
Methods for fixing whole organs, for sectioning tissues before staining, and for histochemical staining of organs and sections have been described (Lin et al. 1990a,b,1992). The studies of Lin et al. (1992) are particularly notable because they describe alternative staining reactions yielding different-colored products for each histochemical marker enzyme. They also describe the most effective combination of double staining when mixtures of two differently tagged tumor cell types are being analyzed in the same tissues. In this latter case, the order of fixation, inactivation of host enzyme activities, and staining reactions become critical to maximize color resolution. Research Organics, Inc. (Cleveland, OH) have generated a large array of histochemical substrates for β-galactosidase, alkaline phosphatase, and alcohol dehydrogenase enzymes whose products have very different colors and are easily differentiated.
Earliest Events in Fibrosarcoma Primary Tumors
Little is known in any experimental animal model system about which events are critical for developing primary tumors at any injection site of recipient animals. Because of the ultrasensitivity of detection of single tumor cells tagged with histochemical marker genes, we undertook analyses of the multicellular organization and expansion events hours or days after sc injection of LZEJ or APSI tumor cells (both are tumorigenic but only the former spontaneously metastatic from this site) (O'Connor and Culp 1994).
As shown in Figure 1 in a time-course experiment, either ras- or sis-transformed 3T3 cells behave virtually identically in terms of multicellular reorganization. Both cell types yielded blue histochemical product with either the IacZ or the ALP marker gene, based on substrate selections described previously (Lin et al. 1992). Cells yield a “curly-haired” organization pattern within 1 hr after injection along the paths of least resistance in the subcutis. This is followed by loss of cells at the periphery and by condensation/coalescence of tumor cells into a more concentrated bolus by 6 hr. By 24 hr, both cell classes are showing migratory behavior at the peripheries of the bolus. By 6 days there are differences between the two cell types, with the ras transformant growing out more effectively and cells aligning into spreading patterns. This was not so apparent with the sis transformant. When this experiment was repeated with IacZ-tagged 3T3 cells (un-transformed), cell clearance proceded rapidly by 24 hr and was virtually complete by 6 days (O'Connor and Culp 1994). Therefore, the oncogenes have conveyed significant survivability of transformed cells in the subcutis that permits tumor outgrowth after several days of stabilization.
This system provided even greater detail at higher magnification in terms of single-cell responses (see Figure 3 in O'Connor and Culp 1994). Sis-transformed cells failed to spread their cytoplasm very effectively at all time points in the subcutis. In contrast, ras-transformed cells spread quite effectively by 6 hr and this spreading undoubtedly led to the “spreading pattern” of these cells (Figure 1H) that is not apparent with sis transformants (Figure 1G).
Ultrasensitive luminometry assays for β-galactosi-dase and alkaline phosphatase activities were then developed to quantitate these earliest events in primary tumor development (O'Connor and Culp 1994). During the first 24 hr, 30–60% of the transformed cells are cleared from the subcutis, and clearance of 3T3 is greater than 90%. Stabilization then occurs for the transformants but not for 3T3, followed by cell number expansion for the transformants by 96 hr, particularly for the ras transformant.
These studies revealed several important facets of the initiation of primary tumors. First, oncogene transformation induces a mechanism(s) of cellular resistance to the clearing mechanism, possibly mediated by NK cells or some other circulating immunoregulatory cell type (Heppner and Miller 1983; Nowell 1986; Barker and Reisfeld 1993; Nicolson 1993). Alternatively, some clearance may be mediated by apoptosis induced in cells that are incapable of finding a suitable matrix on which to settle. 3T3 cells were very susceptible to clearance and were the most fastidious with regard to their relationship with the extracellular matrix. Whether this resistance to clearance is based on tumor cell adhesion alterations with the extracellular matrix and/or on release of some soluble factor from the tumor cells remains to be determined.
The second finding relates to the first above, i.e., ras-transformed cells interact with the extracellular matrix by mechanisms that differ from their parental 3T3 cells or from sis-transformed cells. Both transformants yield a bipolar spreading morphology in culture. In vivo, LZEJ cells spread effectively on the matrix of the subcutis, whereas APSI cells remain rounded. These results indicate the very different nature of matrix adhesion mechanisms operating in the culture dish vs. those in the complex animal tissue. Because extracellular matrix adhesion plays a significant role in signaling into the cell to elicit various physiological and genetic events (Juliano and Haskill 1993; Clarke and Brugge 1995), these differences may indicate a primary mechanism for altered clearance of these cell classes as described above.
A third major finding of this study is the expansion of the two tumor cell populations. In contrast to the hypothesis that only one or a few tumor cells divide to become the predominant cell type in the primary tumor, these studies clearly show that many cells in the subcutis-implanted population divide and expand to form the primary tumor. This heterogeneity and diversity of dividing cells undoubtedly lead to the diversity of tumor cell subsets observed in larger primary tumors (Heppner and Miller 1983).
Finally, these studies reveal that clearance of some tumor cells does occur quantitatively, and the qualitative studies (staining of the subcutis) indicate that dispersed and low-density tumor cells at the periphery of the inoculated mass are the most vunerable population. They also suggest that survivability depends to some extent on tumor cell density and/or numbers. Either the killing mechanism becomes saturated out or it becomes unable to take on large concentrations of tumor cells (possibly because the latter release a “neutralizing” factor into their environment, whereas 3T3 cells are incompetent in this regard). Histochemically tagged tumor cells will now enable us to evaluate factors and gene activities at the single-cell level that are critical for these initiating relationships for primary tumors.
Earliest Events in Neuroblastoma Primary Tumors
In parallel, we investigated human neuroblastoma cells for their tumor development characteristics at two ectopic sites, subcutaneous and intradermal (Kleinman et al. 1994), as well as at an orthotopic site (the adrenal gland, because this would be one site of origin of human neuroblastoma). Human neuroblastoma cells were shown to be metastatic from the adrenal gland but not from either ectopic site (Flickinger et al. 1994). We evaluated the earliest events of primary tumors at these ectopic sites in an attempt to resolve any cell type or tumor type specificity in these processes (Kleinman et al. 1994), compared to fibrosar-coma studies described above. Human Platt neuroblastoma cells were tagged with the IacZ marker gene for these studies.
At 1 hr in either SC or intradermal sites, cells distributed into “curly-haired” groups similar to those seen with fibrosarcoma (Figures 2A1 and 2A2 ). By 48 hr (Figures 2B1 and 2B2 ), these collections had condensed into dense ovoid masses that were expanding by 1 week (Figures 2C1 and 2C2 ). At 3 weeks, just as tumors were becoming palpable, large ovoid masses had developed further, most of which retained histochemical stainability when two different clones of transfectants were used (Figure 2D 1), and a third clone variably lost expressibility of the IacZ gene (e.g., Figure 2D 2). These analyses also demonstrated that many tumor cells, not merely a few of them, were dividing to form the primary tumor mass at these earliest time points. The topological relationships between tumor cell collections and neighboring blood vessels is particularly striking at the intradermal injection site (series 2 micrographs of Figure 2). Two other independent IacZ transfectants of Platt cells behaved identically to this clone 3 (Kleinman et al. 1994).
Our ability to detect blood vessels that retain red-staining erythrocytes after transcardial perfusion of fixative encouraged us to examine when blood vessels were developing proximal to the tumor cells themselves (Figure 5 in Kleinman et al. 1994). By 48 hr at the intradermal site, very small blood vessels were detectable branching from large vessels and in the direction of the tumor cells. That these “microvessels” were not seen in neighboring regions of the tissue lacking tumor cells is suggestive of angiogenic/chemotactic responses specifically to tumor cell-secreted factors. By 2 weeks of primary tumor development, major blood vessels had grown directly into the tumor mass.
Therefore, these studies demonstrated that a very different tumor class, neuroblastoma, developed primary tumors with very similar organizational patterns to those observed with fibrosarcoma at SC or intradermal injection sites. Furthermore, they demonstrate the high resolution available at the intradermal site for detecting new blood vessel development in response to tumor cells, with the former staining red and the marker gene allowing detection of blue-staining tumor cells. Although the three IacZ transfectants of Platt spread in culture and in some cases developed neurites, they remained rounded at all three injection sites (SC, intradermal, and adrenal), indicating again that extracellular matrix adhesion relationships of tumor cells in vivo are quite different from their responses in vitro (Albelda et al. 1990). It is also clear from these studies and those above that histochemically tagged tumor cells provide the resolution required to evaluate specific cell and molecular events operating for or against cell survival and division in various host tissue environments.
Micrometastasis to Lung: Significance of Microfoci
Lin et al. (1990a,b) reported the ultrasensitivity of detecting tumor cells metastasizing into the lungs of athymic nude mice via spontaneous processes from the sc injection site for fibrosarcoma cells or via experimental metastasis by tail vein injection. This system was explored further by a variety of experimental approaches.
One of the real surprises of these first studies (Lin et al. 1990b), using an experimental metastasis model in nude mice, was the high frequency of multicell foci that were observed in the lungs within minutes after tail vein injection. This contrasted with the single-cell distribution of the same cells inoculated into culture and indicated that the accumulation of microfoci of two to seven cells occurred at the microvessel site itself or during the first minute(s) of circulation in the blood.
Multicellular reorganization of fibrosarcoma tumor cells in the subcutaneous site at the earliest time points. APSI (
Lin and Culp (1992a) investigated this mechanism further by pretreating the LZEJ cells with various procedures (fixation, irradiation with 60Co, or mitomycin C) in an effort to promote single-cell implantation in the lung or to further amplify cell aggregation. In all cases, IacZ stainability was retained by the cells for several days after treatment even though cells failed to divide. Figure 3 illustrates some of these findings. The multi-cellular nature of live-cell foci is readily apparent in whole organs (Figure 3A) or in sections (Figure 3E) with X-gal staining. Formaldehyde/glutaraldehye-fixed cells yielded foci that were again multicellular but were much more rounded in overall shape (Figures 3B and 3F). In contrast, 60Co-treated cells yielded foci that were very similar to living cells (Figures 3C and 3G). The same applied to mitomycin-treated cells (Figures 3D and 3H).
Time course of neuroblastoma primary tumor formation at two sites. LZPt-3 cells (105) were injected into either the SC (series 1 micrographs) or intradermal (series 2 micrographs) sites. These sites were identified with two India ink spots on the skin of nude mice, with injection between the two spots. Mice were sacrificed at the indicated times and the skin sites harvested for fixation and histochemical staining with X-gal. (
Whole-organ or section analyses of LZEJ cells after IV injection of live or treated cells. (
Of particular note is that clearance from the lungs was virtually complete within 2–3 days for irradiated and mitomycin-treated cells, whereas fixed cells were not cleared as efficiently and a sizable fraction of live cells persisted throughout. This indicates that clearance of tumor cells from the lungs requires deformability and/or cell surface functions that are altered by fixation but not by irradiation or mitomycin treatments that target DNA metabolism. However, that all treated cell populations were eventually cleared (live cells were not) indicates that cell division, migration, and/or select gene expression patterns are critical for persistence in the lung microarchitecture. When live cells were premixed with fixed cells before tail vein injection, these mixtures yielded microfoci that were completely cleared from lungs (Figure 4 in Lin and Culp 1992a); live cells did not survive in this situation. In contrast, preinjecting live cells, followed 6 hr later with fixed cells, yielded the normal (and stable) subset of micrometastases in lungs. These experiments indicate that fixed cells invoke some dominant mechanism of clearance that live cells cannot overcome.
These findings in the experimental metastasis model system, combined with our development of multiple histochemical marker genes (Lin and Culp 1991, 1992b), prompted us to examine the fates of two different (but related) tumor cell types tagged with different histochemical marker genes. For these studies we compared LZEJ cells with APSI cells, because both were 3T3 derivatives transformed with two different oncogenes (ras vs sis) (Lin et al. 1993). This dual-cell system also tested the hypothesis that nonmetastatic APSI cells that produce PDGF would facilitate certain aspects of the metastatic development of LZEJ cells.
Double staining protocols developed by Lin et al. (1992) were quite effective at resolving the two tumor cell classes in lungs after tail vein injection (Lin et al. 1993). These studies demonstrated that APSI cells (a sis transformant) developed multicell foci when injected into tail veins of nude mice that were virtually identical to those observed with LZEJ cells. When equal numbers of APSI and LZEJ cells were mixed and then inoculated, three classes of lung microfoci were observed by 1 hr (Figure 4): APSI only, LZEJ only, and foci containing both cell types. The mixed foci were more evident at higher magnifications (Figures 4B and 4C) and were verified as mixed foci by analyzing sections of lung for histochemical markers (see Figure 5 in Lin et al. 1993).
Homotypic and heterotypic co-distributions of APSI and LZEJ cells co-injected IV. Lungs were removed from nude mice given an IV injection of a mixture of 105 LZEJ and 105 APSI cells, 1 hr post injection. Lungs were fixed and stained with Red-gal substrate first to detect LZEJ cells (red-staining), then heat-treated, and finally stained with X-phosphate to detect APSI cells (blue-staining). Red-stained LZEJ foci, well isolated from APSI foci, are indicated by black arrowheads; blue-stained APSI foci, well isolated from LZEJ foci, are indicated by black arrows. Double stained foci containing both tumor cell types are also observed at a significant frequency and are indicated by open arrows. From Lin et al. (1993), with permission. Bar = 100 μm.
When these three classes of foci were quantitated with time (Lin et al. 1993), it was clear that the mixed-cell foci were enriched relative to the single-cell metastases, indicating that cooperativity between these two cell types provides some selective advantage. When larger metastases were evaluated in these animals at 3 weeks, all three tumor classes were identified, with approximately one third of each class (Figure 6 and Table II in Lin et al. 1993). Of particular note is that APSI cells, when injected by themselves, rarely developed into large metastases in the lung. When they were co-injected with LZEJ cells, they readily developed into large experimental metastases, suggesting that the LZEJ cells were providing some “factor” or environmental influence to promote APSI growth in this foreign organ. These studies showed that both cell types could develop into specific large metastases (therefore biclonal) and that this system could be used with sufficient resolution to test two different tumor classes that are potentially complementing for progression in various target organs.
Asymmetric Distribution of Two Genetic Classes in Large Primary Tumors
By tagging APSI or LZEJ cells with different histochemical marker genes (Lin et al. 1993), we were afforded the opportunity to evaluate the distribution of these two genetically different (but related) tumor cell classes as primary tumors expanded into large masses at the SC site. Evaluation of more than a dozen tumors that were larger than 7 mm in diameter revealed the same pattern of distribution of these two cell types. They were not co-distributed with each other throughout the tumor at the single-cell level. Rather, they were regionally concentrated into large segments—LZEJ cells only in one segment with APSI cells only in a neighboring segment. In no case did we observe effective intermixing of the two tumor cell classes. This important result indicates regional specialization for genetically different tumor cell classes, as if each cell class preferred its own cell types as neighbors.
There are three general hypotheses as to why regional specialization occurred in this important experiment. First, there may be cell:cell adhesion mechanisms that differ between APSI and LZEJ cells and that promote homotypic intercellular aggregation of each cell type as division ensues to lead to the large tumor. Second, each cell type may secrete growth factors or other factors that specifically promote division of that particular cell type, thereby leading to homogeneous regional concentrations of each cell type. Third, at some point during the first week of primary tumor formation, clonal dominance may become operative and single cells in specific regions may outgrow neighbors, thereby leading to regional concentrations observed later. Genetic and cell biological analyses of cells in each region will be required to resolve these various hypotheses. However, they do shed some light on how genetic variants may arise during expansion of the primary tumor. Cells isolated from specific regions can now be mixed with cells from other regions and tagged differently to evaluate the genetic and pheno-typic bases of this diversity.
Summary and Perspectives
These studies illustrate the power and diversity of histochemical marker gene-tagged tumor cells to analyze the earliest stages of primary tumor formation, as well as micrometastasis formation, in virtually any organ. The use of two different marker genes, yielding differently colored reaction products, facilitates our understanding of how two cell types (or classes) do or do not complement each other in these events. In some cases, synergism was noted between two related tumor classes, e.g., stabilization in the lung of experimental metastases with outgrowth of both classes where usually only one class would grow into overt metastases. Conversely, the two classes may be exclusionary, e.g., regional specialization in large primary tumors of APSI or LZEJ cells without any intermixing of the two classes. This dual-marker gene system can be effective for analyzing other recently discovered concepts in tumor progression. Two examples from our recent studies of fibrosarcoma and neuroblastoma will illustrate the future utility of these marker genes.
Nicolson (1993) first suggested the possibility that some genes are specifically turned on in a subset of primary tumor cells to permit their metastasis to target organs and then are turned off because they are selected against and/or are counterproductive for outgrowth of primary tumors or overt metastases. Our laboratory has now identified one such gene product, CD44, a family of cell surface molecules, some of whose alternatively spliced products can bind hyaluronan (HA). Kogerman et al. (1997a-c) discovered in the fibrosarcoma system (a) that overexpression of the HA binding standard isoform of CD44 promotes micrometastasis to the lung in both spontaneous and experimental metastasis models, (b) that this overexpression is selected against and is downregulated during primary tumor or metastases outgrowth processes by an epigenetic mechanism, and (c) that selective advantage for micrometastasis by CD44 overexpression is mediated by increased implantation/stabilization in the microvasculature of the lung. It will now be possible to tag CD44-overexpressing cells with one marker gene and base-level expressors with a second marker gene, to mix the two cell types, and to introduce them by various routes into nude mice to more carefully evaluate the cell selection and genetic regulation that occurs specifically in vivo.
A second example derives from neuroblastoma tumor progression in which the N-myc oncogene is highly amplified, providing aggressive metastatic capabilities to this tumor (Brodeur 1991). N-myc amplification leads to cells in culture that are rounded and have greatly reduced levels of the integrins α2β1 and α3β1 that recognize extracellular matrices (Flickinger et al. 1994). Evaluation of this downregulation of integrin expression by transfected/overexpressed N-myc oncogene in neuroblastoma cells (in which N-myc was not overexpressed originally) demonstrated two mechanisms of regulation (Judware and Culp 1995,1997a). First, there was transcriptional downregulation of the mRNAs for both α2 and α3 integrin subunits in N-myc-overexpressing cells (either naturally occurring or transfected). Second, this loss of these two prominent α-sub-units led to inability of the β1-subunit to mature during its biosynthesis (possibly because it lacked an α-subunit to complex with) and its eventual degradation in cells. Similar mechanisms were discovered in human osteosarcoma cells into which the overexpressing N-myc oncogene was transfected (Judware and Culp 1997b). These N-myc transfected/overexpressing cells can now be tagged with one histochemical marker gene while a neuroblastoma cell with basal N-myc can be tagged with a second marker gene. The two cell types can be mixed and evaluated for their progression characteristics at both ectopic and orthotopic injection sites (Flickinger et al. 1994). Clearly, marker genes have greatly improved the resolution of phenotypic and genotypic analyses of mixed tumor cell populations at many in vivo sites of progression.
Footnotes
Acknowledgements
Supported in part by NIH research grants (to LAC) CA27755 and NS17139. NRK was supported by the Animal Resource Center of Case Western Reserve University; KLO was a trainee of NIH training grant AG00105.
Particular appreciation is extended to Drs Thomas and Theresa Pretlow, Department of Pathology, who provided invaluable advice and methodologies for pathological analyses of histochemically-stained tissues and sections. We recognize the excellent assistance of Elizabeth Zborowska with the nude mouse animal protocols and some support mechanisms from the Cancer Center at Case Western Reserve University School of Medicine (P30 CA43703).