Overexpressed human CD44s promotes lung colonization during micrometastasis of murine fibrosarcoma cells: Facilitated retention in the lung vasculature

Abstract

Normally nonmetastatic murine sis-transformed BALB/c 3T3 cells, transfected with human CD44s gene (hCD44s), acquire spontaneous metastatic capacity to the lung. The mechanism(s) of this facilitated micrometastasis was analyzed in an experimental metastasis model. Human CD44s overexpression promoted the earliest stages severalfold (initial implantation and subsequent stabilization of tumor cells) but was irrelevant for later stages (subsequent outgrowth) of lung experimental micrometastasis. By injecting mixed populations of parental (nonmetastatic) and CD44s-transfected cells, it was shown that cell–cell adhesion between tumor and parental cells was not promoted by hCD44s but that promotion of cell–cell adhesion to lung endothelium or specifically between transfected cells (via hyaluronan) are likely mechanisms. Results obtained with hCD44s-negative primary tumor cells and hCD44s-positive or -negative variants of lung micrometastatic cells (after s.c. injection of transfectants) confirmed the importance of CD44s overexpression for early but not late stages of experimental lung metastasis. Therefore, CD44s represents a metastasis-facilitating molecule that is irrelevant for primary tumor outgrowth but that promotes micrometastasis to the lungs at the very earliest stages.

Metastasis is a complex cascade of interrelated events (1), successful completion of which depends on the intrinsic variability of tumor cells (2) and host microenvironments (3, 4). This complexity has made the molecular mechanisms of metastasis difficult to analyze. A number of important molecules have been suggested, including CD44, a cell surface receptor for the high molecular weight glycosaminoglycan hyaluronan (HA) (5).

CD44 is a heterogeneous family of type I transmembrane glycoproteins with complex alternative-splicing and glycosylation patterns (6). Interest in CD44 was generated by reports implicating CD44 variant and standard (CD44s) isoforms in tumor metastasis in animal models (5, 7). Numerous studies of CD44 isoforms as metastasis markers in human cancers, such as colorectal carcinoma (8–10) and breast cancer (11–14), produced variable results.

Studies from our laboratory correlated serum-independent CD44s expression with tumor metastasis in a murine fibrosarcoma/metastasis model system [i.e., BALB/c 3T3 cells transformed with sis or ras oncogenes (15–17)]. The involvement of CD44s in metastasis was tested more directly by transfecting the human CD44s gene (hCD44s) into two different nonmetastatic fibrosarcoma cell types. They demonstrated induction of spontaneous metastasis by overexpression of hCD44s (18). In contrast, hCD44s overexpression was selected against during primary tumor outgrowth; this expression was down-regulated in primary tumors by an epigenetic mechanism involving methylation of the human CD44s gene.

In this study, the mechanism(s) of induction of metastatic capacity by hCD44s expression was analyzed using tail vein injections into nude mice and quantitation of pulmonary micrometastases by virtue of the drug resistance markers of the tumor cells. By analyzing both early and late stages of lung colonization, our results show that hCD44s overexpression strongly facilitates the retention of tumor cells in the pulmonary vasculature and their subsequent colonization of the lung.

MATERIALS AND METHODS

Cell Lines for This Study.

Cell lines have been described (15–18). In brief, c-sis-transformed BALB/c 3T3 cells (HUSI, the parental cells) were transfected with an expression plasmid containing the hCD44s cDNA. Two stable transfectants (clones 5 and 6) from independent transfections expressing hCD44s on the cell surface were chosen. After s.c. injection of transfectants into nude mice, several primary and lung metastatic tumor cell lines were isolated and are used herein: 5.2, primary tumor cells from transfectant 5, negative for hCD44s expression; 6.5.1, lung metastatic cells from transfectant 6, hCD44s-positive; and 6.5.2, a variant of 6.5.1 that has lost hCD44s expression. Cells were grown in DMEM plus 10% newborn calf serum (GIBCO/BRL).

Experimental Metastasis and Colony Growth Assays.

Transfected/transformed or tumor cells were detached from tissue culture plates by brief trypsinization, and complete medium was added to inhibit trypsin. Cells were rinsed with PBS and resuspended in PBS at 10⁶ cells/ml. Cell suspensions (10⁵ cells in 100 μl) were injected into tail veins of female athymic nude mice (6–8 weeks old with three mice per datum point unless specified otherwise). Mice were killed postinjection (1 h, 24 h, or 4 weeks). Lungs were dispersed into culture with 0.25% (wt/vol) trypsin/50 mM EDTA/PBS for 45 min at 37°C with shaking. Cells from each lung were plated in complete medium onto four tissue culture plates; 48 h later, drug selection (either 200 μg/ml hygromycin B or 3 μg/ml puromycin) was initiated. When drug-resistant colonies grew out (≈2 weeks after the initiation of selection), two plates per lung were stained with Coomassie blue, and colonies were enumerated; the rest of the cells was used for fluorescence-activated cell sorter (FACS) analyses (see below). The number of tumor cells in lung was determined by multiplying the number of colonies on a plate after selection by the number of plates. For mixed-population experiments, the number of hCD44s transfectant cells was determined as n₁ = n of puromycin-resistant colonies. The number of parental HUSI cells was established using the following formula: n₂ = n of hygromycin-resistant colonies − n₁.

FACS Analysis to Evaluate Cell Surface CD44 and HA Binding.

Cells were collected from culture plates by treatment with 5 mM EDTA in PBS. Cells were resuspended in stain-wash medium [SWM; PBS/0.5% (wt/vol) BSA/0.01% (wt/vol) NaN₃] at 2–5 × 10⁵ cells/ml. Hybridoma supernatant [50 μl of A.3 or 7.10, specific for human CD44 (19, 20)] was added to 50 μl of cell suspension. After a 1-h incubation on ice, cells were rinsed with cold SWM, and 50 μl of fluorescein isothiocyanate conjugate of a secondary antibody (1:40 diluted in SWM) was added to the cells for 1 h. Cells were then washed four times in SWM and analyzed on a FACScan flow cytometer (Becton Dickinson). Negative controls included using normal mouse immunoglobulin G instead of primary antibody and staining parental HUSI cells with A.3 or 7.10 antibody.

FITC-conjugated high molecular weight HA (FITC-HA) has been described (19). It was dissolved in SWM and added to cells for 1 h. After four washes in SWM, cells were analyzed on FACScan. Controls included preincubation of cells with unconjugated HA or with antibodies specific for human CD44 (7.10).

RESULTS

hCD44s Overexpression Facilitates Lung Colonization of Tumor Cells After i.v. Injection.

hCD44s overexpression induces lung micrometastasis formation in normally nonmetastatic c-sis-transformed 3T3 cells (HUSI; ref. 18) in two independent isolates (transfectants 5 and 6). Analysis suggested that hCD44s overexpression was important after the tumor cells had entered the bloodstream (data not shown). To test this possibility directly, experimental metastasis assays were performed comparing parental HUSI cells and hCD44s transfectant 5. Cells were injected into the tail veins of athymic nude mice; at different times postinjection (1 h, 24 h, or 4 weeks), mice were killed, their lungs were dispersed, and cells were plated onto 100-mm tissue culture plates. Tumor cells were quantitated using a colony growth assay based on the drug resistance phenotype of tumor cells. After hygromycin B selection, plates were stained with Coomassie blue, and colonies were counted.

Fig. 1A illustrates colony growth with quantitation in Fig. 1B. There is a significant increase in the number of cells recovered from lungs with hCD44s transfectant 5 relative to parental HUSI cells as soon as 1 h after injection (3-fold increase; Fig. 1B). The differences are even more pronounced at the 24-h time point (9- to 10-fold), indicating that, in addition to facilitating initial implantation, hCD44s may be as important at preventing clearance of tumor cells and/or their “establishment” in the tissue. A second isolate of hCD44s transfectants (#6) demonstrated this statistically significant increase as well (768 ± 58 vs. 61 ± 5 for parental HUSI cells; Fig. 1B).

Figure 1.

hCD44s overexpression promotes early stages of lung micrometastasis formation. Experimental metastasis assay was performed as described in Materials and Methods. In brief, 10⁵ tumor cells were injected into tail veins of athymic nude mice (three mice per data point except for 2 mice for the 4-week time point). At indicated times postinjection, the mice were killed, their lungs were dispersed into culture, and tumor cells were quantitated in a colony growth assay in selective media containing hygromycin B. The colonies were visualized by staining with Coomassie brilliant blue. Transfectants 5 or 6, two independent isolates of hCD44s-transfected HUSI cells. (A) shows representative examples of stained plates from each data point. Quantitation of data is given in B. Statistically significant differences between lung colonization by parental HUSI and transfectant cells: ∗, P < 0.05; ∗∗, P < 0.005; §, these groups consisted of two mice.

Lung-Colonizing Cells Have Elevated Cell Surface hCD44s and HA Binding.

Selective pressures for hCD44s overexpression during lung colonization were analyzed next by comparing CD44 levels in cells recovered from the lungs vs. cells maintained in continuous culture for the same time period. hCD44s transfectant cells tended to lose their hCD44s expression when maintained in tissue culture (Fig. 2A). In Fig. 2B, enrichment for hCD44s-expressing cells as a selected subpopulation is evident from lungs 1 h postinjection. This enrichment is greater at the 24-h time point (Fig. 2C). Although cells from lungs 4 weeks postinjection contained more hCD44s protein than cells in culture for the same length of time (Fig. 2D), this difference was less pronounced than that from cells at earlier time points. “Recovered” cells eventually lost their hCD44s expression upon longer term culturing, suggesting that selection for hCD44s overexpression in micrometastatic selectants was extremely strong, enabling us to detect it even after culture outgrowth and drug selection over a 2-week period. Therefore, hCD44s expression is most beneficial for the earliest stages of microvessel establishment/extravasation but may not be beneficial for later metastasis outgrowth.

Figure 2.

hCD44s overexpression is selected for during early stages of micrometastasis. FACS was performed as described in Materials and Methods. In brief, nonpermeablized cells were incubated with control (mouse immunoglobulin G, negative control) or anti-human CD44 (A.3) antibodies followed by incubation with FITC-conjugated anti-mouse antibodies. Fluorescence of stained populations was analyzed on a FACScan flow cytometer; cell fluorescence histograms are shown here. (A) Injected, A.3, the transfectant cells before injection (stained with A.3); cultured, A.3, transfectant cells maintained in selective media for 3 weeks (A.3 staining). (B) Micrometastatic cells isolated from the lungs 1 h after injection into tail veins (hygromycin-selected); (C) Tumor cells from the lungs isolated 24 h after injection (hygromycin-selected). (D) Tumor cells isolated from the lungs 4 weeks after injection (hygromycin-selected).

The binding of HA was evaluated for transfectant cells maintained in culture or recovered from the lungs at different time points. Parental HUSI cells do not bind HA (Fig. 3A). In contrast, transfectant cells recovered from lungs at 24 h (Fig. 3B) bind significantly more HA than cells maintained in culture (Fig. 3C) or recovered after 4 weeks of lung growth (data not shown). Furthermore, this binding can be inhibited partially with 7.10 antibody specific for human CD44 (Fig. 3B). These results offer a basis for selection of hCD44s variants during early stages of lung colonization.

Figure 3.

Transfectant cells isolated from the lungs display elevated hCD44s-dependent hyaluronan binding. Cells were stained with FITC-HA with or without preincubation with excess unconjugated HA (negative control) or with anti-human CD44 antibodies (FITC-HA/7.10). FACS was described in Materials and Methods and in the legend to Fig. 2. (A) Nontransfected parental HUSI cells. (B) hCD44s transfectant 5 cells, recovered from lungs 24 h after tail vein injection and hygromycin-selected. (C) hCD44s transfectant 5 cells maintained in hygromycin-containing media for 3 weeks.

hCD44s-Mediated Cell–Cell Adhesion Specific for Transfected Cells and/or with the Lung Endothelium as Likely Mechanisms of Action.

Two mechanisms for hCD44s may be envisioned: (i) hCD44s could facilitate cell–cell adhesion with increased retention of cell aggregates in lung vasculature and/or (ii) it promotes adhesion of cells to endothelial cells and/or basement membrane. If heterotypic cell–cell interaction was critical, two predictions can be made: (i) metastasis of transfectants would be decreased when coinjected with parental cells and (ii) selection for hCD44s overexpressors would be stronger in the resulting lung metastatic populations. The exception would be if transfected cells adhere to each other specifically via the overexpressed form of hCD44s, perhaps via HA. Parental HUSI and transfectant 5 cells were mixed at different ratios (80%:20%, 50%:50%, and 20%:80%) and mixtures were injected (10⁵ cells) into tail veins. At 24 h, lungs were dispersed into culture for analysis. Two different selectable markers were used: hygromycin to recover all tumor cells and puromycin to recover hCD44s transfectants only.

Table 1 shows no reduction in the efficiency of lung colonization by hCD44s-expressing cells when coinjected with increasing numbers of parental HUSI cells. These data also show that parental cells formed 40 ± 29 colonies per 10⁵ injected cells at the 24-h time point whereas transfectants formed 764 ± 131. This difference is highly significant when analyzed statistically by a two-tailed heteroscedastic t test (P = 6.4 × 10⁻¹⁰). Therefore, hCD44s may not function by inducing cell–cell adhesion between tumor cells and parental cells but may do so specifically between transfected cells (cell–cell adhesion mediated by HA, perhaps) or via interactions with the lung microvasculature.

Table 1.

Micrometastasis of hCD44s transfectants is not inhibited by nontransfected HUSI cells

Colonies at 24 h derived from:	Injection: HUSI cells/n of transfectant 5 cells, n
	80,000/20,000		50,000/50,000		20,000/80,000
	HUSI	transfectant 5	HUSI	transfectant 5	HUSI	transfectant 5
Actual numbers	36 ± 8	137  ±  30	31 ± 16	399  ±  31	7 ± 12	645  ±  60
per 105 cells, n	45 ± 10	687  ±  150	61 ± 32	797  ±  62	33 ± 58	807  ±  75

Experimental metastasis assays were performed as described in Materials and Methods and in the legend to Fig. 1. Number of transfectant-derived colonies was determined by puromycin selection; number of HUSI-derived colonies was determined by subtracting the number of puromycin-resistant colonies from the number of hygromycin-resistant colonies. 

Cell surface hCD44s protein levels were analyzed next in populations of tumor cells recovered from lungs of animals injected with different ratios of parental cells and hCD44s transfectants. Fig. 4 shows very similar levels of hCD44s in transfectant cell populations from animals receiving mixtures of parental and transfectant cells relative to populations from injection of pure transfectant cells (Fig. 2C). These findings support the conclusions above.

Figure 4.

No increased selection for hCD44s overexpression in micrometastatic transfectants injected in the presence of excess nontransfected cells. Parental HUSI cells were mixed with hCD44s transfectant 5 cells at the ratios of 80%:20% (H80:20), 50%:50% (H50:50), or 20%:80% (H20:80), and the mixtures were injected into tail veins. At 24 h postinjection, lungs were harvested, and transfectant-derived micrometastatic cells were recovered by selection in puromycin. Cells were stained with negative control or A.3 antibodies and analyzed by FACS. Cell fluorescence histograms are shown here.

Experimental Metastasis of Primary Tumor and Lung Micrometastatic Cells Derived from hCD44s Transfectants.

hCD44s-negative primary tumor and hCD44s-positive lung micrometastatic cell lines were isolated after s.c. injection of hCD44s transfectants (18). These cells were used to test whether hCD44s expression was required for experimental metastasis of these in vivo-selected tumorigenic variants.

Primary tumor cells (5.2 cells) that had lost hCD44s protein by an epigenetic mechanism were injected i.v. into nude mice. A slight increase was seen in colonization number relative to untransfected cells at 1- and 24-h time points; this increase was not statistically significant (compare 5.2 cells in Fig. 5 with HUSI cells in Fig. 1). The number of colonies was much smaller, however, than the number of colonizers with transfectant cells (compare 5.2 cells with transfectant 5 cells in Fig. 1). Of interest, the tumor cells recovered from the lungs showed some staining for hCD44s protein, suggesting again selection for hCD44s (data not shown). The animals killed at 4 weeks displayed extensive lung metastasis (Fig. 5C). Tumor cells were recovered from the lungs of both animals and were negative for hCD44s protein (data not shown), indicating that this expression was not required for tumor outgrowth.

Figure 5.

hCD44s overexpression strongly promotes early lung colonization by in vivo-selected tumor cells. Experimental metastasis assay was performed as described in Materials and Methods and in the legend to Fig. 1. 5.2 cells, hCD44s-negative primary tumor cells from s.c. injection of hCD44s transfectant 5; 6.5 cells, lung metastatic cell populations from s.c. injection of hCD44s transfectant 6; 6.5.1 cells, hCD44s-positive variant; 6.5.2, hCD44s-negative variant. (A) Examples of Coomassie-stained colony plates from 1- and 24-h data points; quantitation of results is given in B. Statistically significant differences: ∗, P < 0.001; ∗∗, P < 0.005; §, P < 0.01 for the difference with 6.5.1 cells and P < 0.005 for the difference with 5.2 cells. (C) Photographs of lungs isolated 4 weeks after injection of tumor cells; extensive metastasis formation and displacement of normal lung tissue are evident.

Finally, micrometastasis of hCD44s-positive lung micrometastatic cells [after s.c. injection of transfectant 6 (6.5.1 cells)] was tested. These tumor cells express sizable amounts of hCD44s. As shown in Fig. 5, lung colonization efficiency of these cells was similar to that of hCD44s transfectants at 1 h and 24 h; these second-round tumor cells recovered from lungs contained very high levels of hCD44s (data not shown). At 4 weeks, large metastases also were detected (Fig. 5C). Cells from these lungs with late-stage disease contained much lower levels of hCD44s protein (data not shown).

From 6.5.1 cells, a spontaneously arising variant was isolated (6.5.2 cells) that had lost hCD44s expression. When these cells were i.v.-injected (Fig. 5), they formed significantly fewer colonies than 6.5.1 cells at 24 h, further demonstrating the importance for hCD44s in lung colonization. However, these cells were significantly more metastatic than parental HUSI or 5.2 cells. Also, no enrichment for hCD44s expression was seen in tumor cell populations from lungs (data not shown), indicating that these in vivo-selected cells had acquired additional mechanisms allowing their homing to the lungs.

DISCUSSION

These results demonstrate strong enhancement of lung experimental micrometastasis by overexpression of hCD44s in normally nonmetastatic murine fibrosarcoma cells. This enhancement was operative during early stages of lung colonization [at 1-h and especially at 24-h time points when maximal clearance has occurred (21)] and may or may not be dependent on cell–cell interactions of hCD44s-expressing tumor cells as established after injection of mixed populations of cells. One likely explanation is that hCD44s overexpression may enhance adhesion of tumor cells to the lung endothelium and/or their stabilization once penetrating the interstitium. These results were confirmed by comparing the lung micrometastatic capacities of hCD44s-negative primary tumor cells with hCD44s-positive or -negative variants of lung metastatic cells after s.c. injections of hCD44s-transfected fibrosarcoma cells.

In light of our findings, it recently was shown that CD44 and its binding to HA mediate adhesion of lymphocytes to a model endothelium under physiological flow conditions (22). Overexpression of hCD44s in our tumor model also elevated hyaluronan binding; whether this property is also critical for promotion of experimental micrometastasis remains to be tested more directly.

Whereas hCD44s overexpression was selected for during early stages of experimental metastasis, it was irrelevant (or even counterproductive) during the outgrowth of overt metastases. Cells from large lung metastases had very low levels of FACS-detectable cell surface hCD44s. The metastatic phenotype is expressed by a minor subpopulation of cells in primary tumors (23); these same properties may be irrelevant or disadvantageous for tumor outgrowth at a given site and, therefore, be only transiently expressed in these subpopulations (3). Our results suggest that CD44s may be a critical part of such a transient metastatic phenotype.

Studies on the expression of CD44 in human clinical cancer specimens have produced variable results, suggesting that CD44 may not be required for metastasis of some human cancers (24). Our results offer potential explanation for these disparate results, contrasting with the promising studies with animal models (5, 7, 18). If CD44 is important for micrometastasis but irrelevant or disadvantageous for tumor outgrowth as suggested by our results, the variable expression of CD44 would be expected in human clinical tumor specimens that are principally derived from large clinically detectable tumors. Clearly, much more study of gene regulation is required at the earliest stages of malignant progression in human cancer, particularly those that are transient.

ABBREVIATION

HA

hyaluronan

CD44s

CD44 standard isoform

hCD44s

human CD44s

SWM

stain-wash medium

FITC-HA

FITC-conjugated high molecular weight HA

HUSI

parental sis-transformed 3T3 cells