Adult, foetal and transformed fibroblasts display different migratory phenotypes on collagen gels: Evidence for an isoformic transition during foetal development

Normal fibroblasts have been cultured in the laboratory for many years. When grown under conventional tissue-culture conditions, such cells display a common set of behavioural characteristics, which have come to define the normal phenotype in vitro. When diploid fibroblasts or so-called ‘normal’ fibroblastoid cell lines are exposed to oncogenic viruses or other carcinogenic agents, a proportion of the cells subsequently display altered behavioural characteristics (e.g. anchorage-independent growth) and are now said to be transformed. Many such markers of the transformed phenotype have been identified over the past few years (Cameron & Pool, 1981). It is now clear that not one of these aberrant behavioural characteristics is consistently expressed by all transformed cells. Indeed, clones of cells isolated from a given population of transformed cells may display only one, or various combinations, of these transformation markers (Risser & Pollack, 1974). A detailed account of the transformed phenotype and the potential difficulties that may be encountered in assaying for transformation are presented in the excellent review by Cameron & Pool (1981).

Much of the previous work in our laboratory has been concerned with examining cell migration on three-dimensional collagen gels. When plated on the surface of collagen gels, both normal and transformed fibroblasts rapidly begin to migrate down into the three-dimensional collagen matrix (Schor, 1980). Cells within the collagen matrix are clearly distinguishable from those confined to the gel surface and are easily visualized by focussing down through the gel. We have previously described techniques for expressing this movement of cells into the gel matrix in quantitative terms (Schor, 1980).

In this paper we describe a new assay system that makes it possible to distinguish between normal adult fibroblasts, foetal fibroblasts and transformed fibroblasts on the basis of their migratory behaviour on such gels.

Explant cultures were established from 14 samples of foreskin (male children, age range 2-8 years), 24 samples of human foetal skin, and 24 samples of normal adult skin (Ham, 1980). Samples of foetal skin were obtained from induced abortions (estimated 10-16 weeks) kindly supplied by T. Carr (Clinical Research, Paterson Laboratories). The adult fibroblasts were mainly derived from skin specimens obtained from patients undergoing surgery for conditions unrelated to cancer (e.g. hernia) or from biopsy samples taken from normal individuals under local anaesthesia; skin samples from the following sites were used: 10× females, upper chest (age range 21–76 years); 2× female, abdomen (34, 74 years); l× female, postauricular skin; l× female, shoulder (52 years) ; 2× female, forearm (18, 22 years) ; 1× male, chest; 4× male, abdomen (31–61 years); lx male, face (42 years); l× male, back (67 years) and l× male, lung. The following transformed cell lines were used: HT1080 (human fibrosarcoma), CC22 (human colon), CC21 (human colon), EJ (human bladder), CHED6T (spontaneously transformed Chinese hamster embryo), SV3T3 (SV40 virus transformed 3T3), PyBHK (polyoma virus transformed BHK), JRS (Jensen rat sarcoma), NCTC2472 (mouse tumorigenic fibroblasts), C6 (rat glial tumour), MG63 (osteogenic sarcoma), Py3T3 (polyoma virus transformed 3T3 cells), RSV (Rous sarcoma virus transformed Syrian hamster fibroblasts), SV-W138 (SV40 virus transformed W138 cells) and CCL8 (mouse sarcoma). The CC22 and CC21 cells were a gift from Dr A. Kinsella (Paterson Laboratories), the CHED6T cells were a gift from Dr C. Ockey (Paterson Laboratories), the EJ cells were a gift from Dr R. Hastings and all other cell lines were obtained from Flow Laboratories, Irvine, Scotland. The transformed cell lines all displayed anchorage-independent growth in agar or Methocel; certain of these lines were also tumorigenic.

Cloned populations of fibroblasts were obtained from a particular line of foetal skin fibroblasts. This was accomplished by plating 100 cells derived from a confluent stock culture at passage 9 onto 35 mm plastic tissue-culture dishes. Such sparsely plated cultures were incubated in growth medium in the presence of a gas mixture containing 3% O2 and 5% CO2 in nitrogen. The use of reduced oxygen tension was found to improve plating efficiencies to a significant degree, these varying between 12 and 14%. Individual colonies containing between 100 and 200 cells were then trypsinized using cloning rings with an internal diameter of 4 mm, the resultant cell suspensions plated onto 35 mm plastic tissue-culture dishes and these again incubated under reduced oxygen tension until a confluent culture was obtained. Cultures of cloned cells could then be maintained and passaged in 90 mm dishes as described above for uncloned cells. Since stock dishes contained 1·6×10⁴ to 6·4× 10⁴ cells cm⁻² at confluence, we estimate that 21–22 cell doublings were required to achieve this final cell density from the original progenitor cell in the two in vitro passages used for cloning. This additional number of estimated cell doublings must be taken into account when comparing the passage number of uncloned and cloned cell cultures.

Type I collagen was extracted from rat tail tendons and used to make three-dimensional collagen gels as previously described (Schor & Court, 1979). In our standard experimental protocol, cells are plated onto the surface of replicate collagen gel substrata at a low initial density (i.e. 10³ cells cm⁻²) and a high initial density (i.e. 10⁴cells cm⁻²). After 4 days of incubation at 37°C, the percentage of cells present within the three-dimensional collagen matrix was determined by the ‘microscopic method’ described previously (Schor, 1980). Accordingly, cultures are examined in a Leitz Diavert microscope fitted with a photographic graticule defining a rectangular area of known dimensions. The cells on the gel surface and within the collagen matrix that fall within this rectangular area are counted in 10-15 randomly selected fields for each of the replicate cultures. The number of cells found within the collagen matrix may then be expressed as a percentage of the total number of cells present.

Cell migration into the gel matrix is influenced by a number of experimental parameters (Schor, Schor, Winn & Rushton, 1982). The effects of one such parameter, cell density, on the migration of normal human lung fibroblasts (WI38) and their SV40 virus transformed counterparts (SV-WI38) are shown in Fig. 1. Cells were plated at densities between 10³ and 10⁴cm⁻² and the percentage of cells within the collagen matrix was measured after 4 days of incubation. As can be seen in Fig. 1, the migration of the WI38 cells is inversely proportional to cell density, whereas the migration of SV-WI38 cells increases in a proportional manner.

The data presented in Fig. 1 for WI38 cells and their virally transformed counterparts suggest that it might be possible to distinguish between normal and transformed fibroblasts by the CDMI values they express in vitro. In order to test this possibility, we examined the behaviour of 14 strains of normal human foreskin fibroblasts, 24 strains of normal human adult fibroblasts (male and female, obtained from different anatomical sites; see Materials and Methods), 24 strains of human foetal fibroblasts and 15 different transformed fibroblast cell lines. Cells were plated on replicate gels at 10³ and 10⁴ cells cm⁻² and the percentage of cells found within the collagen matrix was measured after 4 days of incubation. The results of over 500 individual experiments are presented in Fig. 2. No difference was apparent between the CDMI values of normal foreskin and adult fibroblasts; in both cases over 90% of the cells examined had CDMI values greater than +0·4. The CDMI values of the transformed cells produced a completely non-overlapping distribution profile; in this group greater than 90% of the cells examined had negative CMDI values. The response of the foetal cells was especially interesting. These cells behaved as a clearly defined group, with CDMI values falling between those of the normal (foreskin and adult) and transformed cells.

The data presented in Fig. 2 indicate that it is possible to define empirically certain ranges of CDMI values, which include the majority of specimens of a given cell type (i.e. normal, foetal or transformed). Such divisions are indicated in Fig. 2 and are defined as follows: the transformed range (T) with CDMI values <—0·4, the transformed/foetal range (T/F) with CDMI values between —0’4 and 0, the foetal/normal range (F/N) with CDMI values between 0 and +0·4 and the normal range (N) with CDMI values >+0·4.

The majority of our experiments were conducted with fibroblasts originally established from explant cultures using small pieces of tissue; such cultures are consequently derived from the relatively small number of cells capable of active migration out from the tissue fragment. In order to determine whether a particular subset of cells (possibly exhibiting a ‘migratory phenotype’) might be selected preferentially in such explant cultures, a number of experiments were performed in which replicate samples of skin were digested with collagenase overnight and the resultant freed cells and tissue debris used to establish parallel fibroblast cultures. Fibroblasts derived from cultures established by either technique (i.e. explant or collagenase digestion) were morphologically indistinguishable, displayed the same kinetics of growth and saturation cell densities and had identical CDMI values (results not shown).

The stability of the CDMI as a function of in vitro passage number is shown in Fig. 3. Data are presented for a normal adult fibroblast, a foetal fibroblast and a human fibrosarcoma cell line (HT1080). Cells were passaged every 7–10 days at a split ratio of 1:5 and the period of study shown in Fig. 3 covers the entire in vitro lifespan of the normal adult and foetal cells. The HT1080 cells are a ‘permanent cell line’ and hence do not exhibit a finite lifespan in vitro’, these cells were, however, studied over a similar number of passages in order to compare their CDMI values directly with those obtained with the diploid cells. The CDMI values of the adult fibroblast strain remained confined to the N range over its entire in vitro lifespan (estimated 74 population doublings), giving a mean value of +0-57 ±0-13 during this time. The CDMI values obtained with the HT1080 cells were similarly consistent, with the majority of determinations falling within the T/F range (mean CDMI =-0·32 ± 012).

The results presented in Fig. 2 suggest that foetal cells undergo a transition at some point in their developmental history, which results in the expression of elevated CDMI values falling in the adult range. That such a transition may also occur in vitro is shown in Fig. 3. Before passage 25 (estimated 58 population doublings), this strain of foetal cells had CDMI values that tended to fall within the F/N range (mean value = +0·05 ± 0·04). We observed an abrupt transition occurring at passage 25, with the cells now expressing CDMI values in the N range. This transition appeared to be a stable attribute of the cell phenotype and continued to be expressed for the remainder of the fibroblast’s in vitro lifespan (mean value = +0·57 ± 0·13). A similar transition of CDMI values from the F/N range to the N range was observed to occur in another population of FS8 cells at passage 24, as well as in another uncloned foetal cell line (results not shown).

Since an uncloned population of foetal cells was used to obtain the data shown in Fig. 3, the observed transition in CDMI values may have resulted from either the preferential growth (or survival) of cells expressing CDMI values in the N range, which existed in the population before passage 25, or from an actual phenotypic transition in the progeny of cells previously expressing CDMI values in the F/N range. In order to examine these possibilities further, 110 clones of cells were obtained from the uncloned strain of foetal cells used to obtain the data presented in Fig. 3 (i.e. strain FS8) ; clones were isolated from the FS8 strain at passage 9 of its in vitro lifespan. A total of 42 clones were subsequently subcultured and CDMI values obtained at passage 14 (i.e. estimated 53 population doublings; N.B. estimated population doubling of cloned cells calculated by adding 21 to estimated population doubling of uncloned cells; see Materials and Methods). As can be seen in Fig. 4A, there is indeed a considerable degree of clonal heterogeneity in CDMI values expressed at this time, with 33 3% of the clones falling within the T/F range, 45-2% within the F/N range and 2T5% within the N range. The range of CDMI values obtained in 36 replicate experiments using one particular clone of cells (FS8/102) is shown in Fig. 4B. A tight modal distribution was obtained in this series of experiments with a mean value of +0’05 ± 0’07. These results suggest that the rather broad distribution profile shown in Fig. 4A represents a true clonal heterogeneity in the population and is not due to experimental error inherent in determining the CDMI value. Since clones of cells were isolated from the uncloned population at one particular in vitro passage number, no information is consequently available regarding the precise time at which cells displaying ‘adult-like’ CDMI values first appeared in the uncloned population.

The CDMI values expressed by 14 randomly selected clones were followed throughout the duration of their in vitro lifespan. Three general patterns of behaviour were observed (Fig. 5). Clones that initially displayed CDMI values falling within the N range invariably continued to do so for the remainder of their in vitro lifespan (pattern A). Clones that initially displayed CDMI values falling within the T/F or F/N ranges either continued to do so until senescence (pattern B), or underwent a well-defined transition to a CDMI value falling within the N range (pattern C). The data obtained from all 14 of these clones are summarized in Table 1. The results indicate that the majority (75%) of clones initially displaying CDMI values in the T/F and F/N ranges underwent a transition to an adult phenotype before becoming senescent. There was no apparent correlation between the initial CDMI expressed by the clones and the estimated number of population doublings at the time of senescence. These data support the view that the transition observed in the uncloned foetal cell population (Fig. 3) is due to a change in phenotype expressed by individual cells and not by the selection of a particular subpopulation.

Transformed fibroblasts are commonly distinguished from their normal counterparts by the expression of certain phenotypic characteristics in vitro, e.g. anchorage-independent growth, focus formation and secretion of elevated quantities of plasminogen activator (Cameron & Pool, 1981). Fibroblasts in vivo are generally embedded within a three-dimensional matrix containing type I collagen. Normal fibroblasts embedded within a three-dimensional collagen matrix in vitro have been shown to exert tension on the collagen fibres and thereby induce the compaction of the gel when it is detached from the walls of the tissue-culture dish (Bell, Ivarsson & Merrill, 1979; Bellows, Melcher & Aubin, 1981; Allen & Schor, 1983). Transformed fibroblasts were subsequently shown to exhibit a considerably reduced capacity for gel compaction (Steinberg, Smith, Colozzo & Pollack, 1980; Buttle & Ehrlich, 1983). The results presented here indicate that normal and transformed cells may also be distinguished from each other by assessing the effects of cell density on migration down into the three-dimensional collagen gel matrix. These data are presented in the form of a single numerical value, the cell density migration index (CDMI).

Although there are many criteria that may now be used to distinguish normal and transformed cells from each other, foetal and adult fibroblasts generally behave in a similar fashion when cultured in vitro and are therefore difficult to distinguish from each other by a simple assay procedure. The differences that we observe in the CDMI values expressed by foetal and adult cells are of particular interest in this context. Foetal fibroblasts have previously been reported to express certain transformation-associated phenotypic characteristics (e.g. anchorage-independent growth) when assayed under the appropriate conditions (Peehl & Stanbridge, 1981). All of the cloned and uncloned foetal fibroblasts examined in our study were indistinguishable from normal (adult and foreskin) cells in terms of their morphology, serum requirements for growth, ability to compact a floating collagen gel, and inability to form colonies in Methocel when studied in the appropriate assay systems (results not shown).

Previous studies have indicated that certain differences in fibroblast behaviour in vitro may be related to the age and sex of the donor, as well as the particular anatomical site of derivation (Conrad, Hart & Chen, 1977; Harper & Grove, 1979; Azzarone & Macieira-Coelho, 1982). Differences in the protein synthetic activity of foreskin fibroblasts and fibroblasts derived from other (non-genital) anatomical sites have recently been demonstrated by Thompson et al. (1983). In contrast, our data suggest that the CDMI value expressed by fibroblasts is not a sex- or site-dependent phenotypic characteristic, since no differences were observed between foreskin fibroblasts and adult fibroblasts derived from a variety of anatomical sites, from both male and female donors.

The underlying mechanisms responsible for the observed effects of cell density on fibroblast migration are not known. Several possible levels of control may be involved, including: (a) cell-produced changes in the organization and/or biochemical composition of the collagen matrix (Steinberg et al. 1980; Schor, Schor & Bazill, 1981; Erickson & Turley, 1983; Yamada, 1983); (b) the production of soluble chemokinetic or chemotactic factors by the cells (Hayashi, Yoshida, Ozaki & Ushijima, 1970); and (c) an effect of local cell density on the social interactions operative between cells (Abercrombie, 1970). These various possibilities are amenable to experimental investigation and will be discussed in a forthcoming article.

It should be emphasized that the CDMI value determined for a given population of cells does not itself provide information concerning the heterogeneity of migratory phenotypes expressed by individual cells within that population. The CDMI is a single numerical value reflecting the average behaviour of cells within the population examined, irrespective of the degree of heterogeneity present. In this context, note that the histograms shown in Figs 2 and 4A depict the distribution of the mean CDMI values expressed by the different uncloned or cloned cell populations examined. The standard deviations of the mean CDMI values obtained for each of the cell populations examined (either cloned or uncloned) were of a similar magnitude (compare data presented in Figs 3, 4B and 5), generally being less than ±0-1. Information regarding the heterogeneity of CDMI values within a given population of cells can be estimated only by examining the behaviour of individual clones of cells derived from that population.

The transition from a foetal to an adult form of haemoglobin has been recognized to occur for some time. Many similar types of ‘isoformic’ transitions have subsequently been observed to occur during embryogenesis and are now believed to play an important role in the coordinate control of normal development (Caplan, Fiszman & Eppenberger, 1983). Our data suggest that foetal fibroblasts undergo such an isoformic transition at some point in their developmental history, which is manifest in vitro by the expression of elevated CDMI values falling in the N range. A similar transition in the in vitro behaviour of foetal fibroblasts during development has been described by Nakano & Ts’o (1981). They reported that clones of fibroblasts that displayed anchorage-independent growth (i.e. a transformation-associated phenotypic characteristic) could be isolated from 10-to 13-day-old Syrian hamster embryos; the number of clones displaying this ‘transformed’ behavioural characteristic decreased as a function of both gestation period and in vitro passage number.

The common occurrence of isoformic transitions allows the interesting hypothesis to be put forward that certain disease states are characterized by the persistent expression of the foetal isoform (of a particular molecule or cell type) in the adult organism (e.g. see Fitzsimmons & Hoh, 1982). Assessment of the CDMI values expressed by fibroblasts in adult patients suffering from different pathological conditions may provide information relevant to this hypothesis.