An immunohistochemical and ultrastructural study of human melanoma colonies grown in soft agar for up to 50 days was performed. Three morphological variants of developing tumor colonies are reported: 1) large light colonies, 2) small dark colonies, and 3) smooth-edged colonies. The large light colony variant is the most frequently observed in the soft agar assay (approximately 70%), followed by the dark colony variant (approximately 27%), and the smooth-edged colony variant (approximately 3%). Major morphological characteristics are associated with each variant, as shown with light microscopy (LM) and transmission electron microscopy (TEM). Both LM and TEM analyses demonstrated that the large light colony variant was hypomelanotic and contained a microfibrillar extracellular matrix (ECM). The small dark colony variant was found to be hypermelanotic and contained a less demonstrable ECM. The smooth-edged variant has an encapsulated periphery, no demonstrable ECM, and tightly packed cells with desmosome-like junctions. In order to characterize further the ECM in the most commonly observed variant, the large light colony, specific antibodies to fibronectin (FN) and collagen types IV and V (COLs IV and V) were applied and observed with immunofluorescence microscopy and immunoperoxidase. In paraffin sections of melanoma colonies, FN was observed associated with both the cell surface and the ECM. However, no specific staining was seen for COLs IV and V. In addition, ruthenium red was used to preserve and selectively bind to glycosaminoglycans (GAGs) and proteoglycans (PGs). TEM studies reveal GAG-like granules stained with ruthenium red in the fibrillar ECM and a dotted, punctate staining of the cell surface. Understanding the biological and architectural composition of developing melanoma tumor colonies in soft agar could contribute to the development of more efficient chemotherapeutic strategies.

In order to quantify the invasiveness of melanoma tumor cells in vitro, a modification of the amniotic basement membrane (BM) model, described by Liotta et al. (Cancer Letters, 11, 141, 1980), was used in combination with radiolabeled tumor cells. B16-F10 metastatic murine melanoma cells and a derived clone (B16-F10L) were prelabeled with 0.1 muCi/ml of [14C]thymidine for 20-24 h in serum-free medium at 37 degrees C. Following incubation, fetal bovine serum was added to a concentration of 5 per cent, and the cells were allowed to grow to confluency for the next 24-28 h. The labeled cells were seeded onto amniotic membranes situated in Membrane Invasion Culture System (MICS) chambers at a density of 2.5 X 10(4) per well. At various times points, radioactivity of tumor cells that completely traversed the membrane was determined using an under-the-membrane sampling method. The average percent invasion demonstrated by the B16-F10 line was 2.75 per cent, and 3.65 per cent exhibited by the B16-F10L cell line after 48-53 h in vitro. Since it was apparent that some variability in thickness existed among membrane samples, a morphological analysis was performed on five sectors of a three-inch-diameter sample from four different placentae. Differences and similarities in BM thickness within the same sector were noted by this technique and could possibly contribute to some variability observed in tumor cell invasion in this model. Another parameter examined was the proliferation of tumor cells in the upper and lower wells of the MICS chambers. By 48 h, approximately 32.1 per cent of the B16-F10 cell line as well as the clone had replicated in the upper wells associated with the BMs compared with a 32.9 per cent replication in the lower wells, which reaffirmed the viability of the tumor cells under experimental conditions and insured similarly replicating populations of cells. In order to quantify the invasiveness of radiolabeled tumor cells accurately through a biological membranous barrier, the proper concentration of cells must be used, tumor cell heterogeneity should be taken into consideration, the technique of sampling radiolabeled invasive cells should be critically analysed, and thickness of the membranous barrier should all be considered as possible important factors in the quantitative analyses.