Effects of the developmental colour mutations silver and recessive spotting on proliferation of diploid and immortal mouse melanocytes in culture

Melanocyte (pigment cell) lines are useful tools for molecular studies of germline coat-colour mutations. After the isolation of normal black mouse melanocyte lines by three groups (Sato et al., 1985; Bennett et al., 1987; Tamura et al., 1987) other lines carrying germline mutations at the b (brown) and c (albino) loci were reported (Abe et al., 1986; Halaban et al., 1988; Bennett et al., 1989). These lines were used to identify the products of the c and b loci: tyrosinase (an enzyme of melanin pigment synthesis) and a tyrosinase-related protein (Yamamoto et al., 1989; Bennett et al., 1990), respectively. They were also used to characterize mutations at these loci with respect to gene sequence (Jackson and Bennett, 1990), mRNA expression (Jackson et al., 1990) and protein properties (Halaban et al., 1988; Halaban and Moellmann, 1990).

Coat-colour in mice is affected by mutations at nearly 60 loci, more than 20 of which have no effect on eye pigmentation and patchy effects on the coat (Silvers, 1979), suggesting that they act on the development of melanocytes in the skin rather than on the biosynthesis of pigment. We are characterizing certain of these mutations, in order to improve our understanding of the genetic control of development and perhaps malignancy. Two such mutations are silver (si) and recessive spotting (rs).

Silver, although known earlier by mouse fanciers, was first described biologically in 1930, as a mutation that reduced the number of pigment granules (melano-somes) of the hairs so that some hairs had few and others none, although some hairs seemed normal (Dunn and Thigpen, 1930). It showed a curious interaction with the brown mutation such that the effect of si was greatest on a B/b background. In other words, when homozygous for si, B/b mice were lighter than either B/B or b/b mice instead of resembling B/B animals as usual (ibid.) Our observations confirm this. Later radiological evidence suggested that si/si skin had reduced numbers of melanocytes rather than a deficiency in the synthesis of melanin (Chase and Rauch, 1950). Quevedo et al. (1981) reported that si/si hair follicles had variable numbers of melanocytes which died prematurely in the hair cycle. Recently a cloned DNA sequence, pmel 17-1, isolated from a human melanocyte cDNA library (Kwon et al., 1987), was found to map at or near the murine si locus (Kwon et al., 1991). Expression of the corresponding RNA was stimulated in melanocytes and melanoma cells by agents that stimulated differentiation (Kwon et al., 1987). However the function of pmel 17-1 is not yet known.

Recessive spotting was first reported by Dickie (1966) as producing head blazes, large belly spots and diluted bellies when homozygous. There has been no published histological description, but one would expect a reduction of melanocyte numbers as observed in other spotting mutants (Silvers, 1979). These effects resembled and interacted with those of heterozygous steel (SI) and dominant white spotting (W^v allele), (ibid.), and rs was mapped close to W with no recombination events so far observed (Southard and Green, 1971; Geissler et al., 1988). It is now known that the W locus encodes the c-kit proto-oncogene, a tyrosine kinase receptor for a novel growth factor encoded at the SI locus and given several names including steel factor (SLF), stem cell factor (SCF) and mast cell growth factor (MGF) (reviewed by Witte, 1990). It has been suggested that the tight cluster of interacting coatcolour loci on mouse chromosome 5, comprising W, rs, Ph and Rw (Silvers, 1979), may have evolved by tandem reduplication and divergence. This is supported by the finding that Ph is a deletion that contains the platelet-derived growth factor receptor A (Pdgfra) gene, which has some homology to Kit (Stephenson et al., 1991); moreover another related tyrosine kinase receptor gene, Flk-1, is closely linked to Kit and Pdgfra (Matthews et al., 1991). A syntenic region is found on human chromosome 4 (Stenman et al., 1989). Thus the rs locus too may encode a tyrosine kinase receptor, whether or not rs is a recessive allele of W or Ph.

Here we describe direct effects of both si and rs on melanocytes in primary culture and on immortal lines after establishment and cloning. Both mutations reduce the cellular profiferation rate.

Tissue culture media and plastics (Nunc) were obtained from Gibco Europe (Uxbridge, UK) and foetal calf serum (FCS) from Tissue Culture Services (Slough, UK) or Flow Laboratories (Irvine, Scotland). Cholera toxin, soybean trypsin inhibitor and 12-O-tetradecanoyl phorbol-13-acetate (TPA) were from Sigma Chemical Co (Poole, UK). TPA and cholera toxin were dissolved and stored as described previously (Bennett et al., 1985), except that cholera toxin stock solution was dialysed to remove azide before storage. The mouse keratinocyte line XB2 (Rheinwald and Green, 1975) was provided by J. Rheinwald and adapted by us to grow without 3T3 feeder cells as previously reported (Bennett et al., 1989).

A supplemented Eagle’s minimal essential medium (SMEM) was used for most purposes. The supplements were penicillin, streptomycin, sodium pyruvate and non-essential amino acids (Kreider et al., 1975) and it was prepared with only 25 mM sodium bicarbonate to give a pH of 6.9 with 10% CO2 (Bennett et al., 1987). Unless otherwise indicated, growth medium was SMEM with FCS at 5% (v/v), and other supplements as specified.

Cultures were prepared from embryos or from mice aged less than 24 hours. “Wild-type” animals (+/+ at the rs and si loci) were either C57BL/6J inbred mice, F₁ hybrids between C57BL/6J and LAC-MF1 mice, or Fj hybrids between C57BL/6J and CBA/Ca mice. The si mutation was studied in animals heterozygous for b (see Introduction), obtained by crossing si/si B/B and si/si b/b mice, si/si B/b and rs/rs mice, both of a C57BL/6J background, were obtained from the Jackson Laboratories, Bar Harbor, Maine. All breeds were maintained at St George’s Hospital Medical School.

Some cultures were prepared as described previously (Mayer and Oddis, 1977; Bennett et al., 1989). More recently the technique was modified to increase the yield of melanocytes and distribute them more homogeneously in culture, as follows. Neonatal mice were killed, washed in 70% ethanol for approximately 10 seconds and then in Dulbecco’s phosphate-buffered saline lacking calcium and magnesium chlorides (PBSA). Trunk skins were incubated with 5mg/ml trypsin in PBSA at 37°C for 1 hour, then transferred to a culture dish containing PBSA. Epidermal sheets were removed, pooled in 100 μl of PBSA with trypsin (250 μg/ml) and EDTA (200 μg/ml) on another plate, and minced finely with a scalpel. The tissue was homogenised in 5 ml SMEM containing 200 μg soybean trypsin inhibitor, by vigorous pipetting with a 2.5 ml Combitip (Eppendorf, Hamburg) and diluted into growth medium (about 10 ml/donor mouse). The suspension was supplemented with cholera toxin (1 nM) and transferred to plates containing mitomycin-treated XB2 keratinocytes as feeder cells. These were plated 1-3 days earlier as described previously (Bennett et al., 1989). TPA (200 nM) was added at the first medium change (1-3 days). Further procedures were as before (ibid.).

Unless otherwise stated, non-immortal melanocytes were grown in SMEM with 5% (v/v) FCS, TPA (200 nM) and cholera toxin (1 nM), while immortal rs/rs and si/si melanocytes were grown in the same medium except with 10% FCS and no cholera toxin. For subculture, cell suspensions were prepared from growing cultures with trypsin (250 μg/ml) and EDTA (200 μg/ml) as described (Bennett et al., 1987), and were replated at 3-5 × 10⁴ cells/ml. Early subcultures were made on to fresh XB2 feeder cells. If cultures became very sparse, Ham’s F10 medium with 18 mM bicarbonate was substituted for SMEM. For cloning, manually selected individual pigmented cells were transferred to separate wells of a 96-microwell plate, each containing XB2 feeder cells and 100 μl growth medium conditioned by them (Bennett et al., 1989).

The culture medium was supplemented with [³H]thymidine (1 μCi/ml) (Amersham) and inosine (25 μM) (Brooks, 1977) for 48 hours. The cells were then fixed with 4% formaldehyde in Dulbecco’s complete PBS for 10 minutes, extracted with icecold 5% (v/v) trichloroacetic acid in water for 1 minute, dehydrated by 3 washes with absolute ethanol and air-dried. Plates were coated with chrome alum-gelatin solution, airdried and coated with Ilford K5 nuclear emulsion (50% v/v in hot water), exposed and desiccated at room temperature for 3 days, and developed.

The plates (usually 3 per treatment) were coded. Labelled and unlabelled nuclei in at least 3 samples of at least 100 pigmented cells were counted per plate, from random positions. The percentage of labelled nuclei was calculated for each plate, and arithmetic means and standard errors for each treatment were determined after decoding.

Crude population doubling times were obtained during routine passage of cell lines, which were not allowed to become confluent. Cells were plated at a known density and doubling times were calculated from the number of days’ growth and the mean cell density attained, as measured by at least 3 independent haemocytometer counts of at least 150 cells each. Melan-a, melan-b and melan-c melanocytes (black, brown and albino) were grown as previously described (Bennett et al., 1987, 1989).

Only unpigmented melanoblasts are initially obtained from neonatal or embryonic mouse epidermis, together with keratinocytes, but melanoblasts become pigmented within about 2 weeks under the given conditions, while keratinocytes are gradually lost. Visual inspection of the resulting cultures by microscopy revealed no obvious differences between wild-type and mutant melanocytes of either type, in their shape, size, migration or differentiation, the last being assessed from the time of appearance of pigment. The appearance of melanocytes of both mutant genotypes is shown in Fig. 1. Their resemblance to other melanocytes was greatest where the latter were growing slowly (Fig. 1C), whereupon the cells became more flattened and heterogeneous, with occasional giant cells even in immortal cultures (Fig. 1A; not shown in Fig. 1C). Giant cells are also common in senescent diploid melanocyte cultures of any genotype.

On further culture of the diploid cells, it became apparent that si/si and rs/rs melanocytes were not reaching confluency as soon as wild-type cultures, suggesting slower growth and/or a higher death rate. Melanocyte proliferation in primary cultures was quantitated by [³H]thymidine autoradiography of pigmented cells. This method was chosen because total cell counts and other measures could be affected by cell death rates or by contaminating non-pigmented cell populations.

In Table 1, melanocyte proliferation in cultures from animals of the three genotypes is compared, in terms of the percentages of melanocytes labelled in 48 hours. Results from cultures of various ages between 3 and 5 weeks (after plating) were pooled from 13 experiments. Cells of all genotypes showed a decline in DNA synthesis with the age of the culture. This decline was at least partially associated with increased cell density; larger colonies showed reduced labelling, especially in their centres where crowding was highest (not shown). However, si/si and rs/rs diploid melanocytes showed this decline at earlier times (and thus lower population densities), as can be seen from Table 1. Overall, the difference in labelling index of melanocytes of both genotypes compared with wild-type cells was highly significant by analysis of variance (Table 1).

Two melanocyte lines were isolated at different times from independent si/si B/b cultures and one from an rs/rs culture; these were designated melan-sil, melan-si2 and melan-rs respectively. The methods used were similar to those described previously (Bennett et al., 1987,1989), though the time taken for the emergence of reliably growing, established cultures was long (8 months for melan-rs, 6 months for melan-sil, not recorded for melan-si2). As before, all three lines retained a requirement for TPA in the medium described, although cholera toxin was omitted once contaminating cell populations were eliminated. Melan-sil and melan-rs were cloned with high efficiency. The slow proliferation from both genotypes continued throughout establishment and at least in early passages subsequently, although growth accelerated at high passage levels as seen with all our melanocyte lines (e.g. Bennett et al., 1987). Doubling times of the lines melan-sil and melan-rs at low passage levels are compared with those of other established melanocyte lines in Fig. 2. Both showed slower population growth than any previous line, with doubling times 3-4 times greater in medium with 10% FCS than melan-a black melanocytes which have the same genetic background.

There may have been a contribution of increased death rate to the increased doubling times, but this cannot have been major, as dead melanocytes can be detected either as floating black cells or by the release of black melanosomes into the medium on cell lysis; and very little death was observed in these lines under the standard conditions. Thus reduced proliferation must have accounted for most of the reduced population growth.

Both melan-sil and melan-si2 cells, when newly established, showed an unusually marked dependence on feeder cells for reattachment and survival after subculture, particularly after frozen storage and thawing, or following more than one subculture without feeder cells (Fig. 3). One day after the illustrated cultures were plated, most of the melanocytes appeared to be unattached and dead in the plates without feeder cells, whereas very few floating melanocytes were seen in those with XB2 cells. This suggested the possibility of alteration in an extracellular attachment (matrix) factor, or a receptor for one. However the effect was lost at later passages, presumably by selection of cells capable of better attachment, and thus could not be characterized further. It may have been a quantitative rather than qualitative difference from other genotypes, perhaps reflecting the general level of health of the cells, as melanocytes of all genotypes grew considerably better in the presence of XB2 cells before immortalization.

Both si/si and rs/rs melanocytes grew more slowly than those of other genotypes, both in primary culture and after immortalization. Both mutations were on the genetic background of C57BL/6J, but the possibility that the slow growth was a result of this background can be ruled out by previous observations that melanocytes from this strain of mice grow well (Sato et al., 1985; Bennett et al., 1987): as well as or better than those from other strains (Tamura et al., 1987). The si/si and rs/rs melanocytes otherwise appeared generally normal, except for a possible deficiency of attachment in si/si melanocytes. These findings are consistent with the postulation that the rs mutation affects a growth-factor receptor, while not proving this. If rs is indeed allelic with W/Kit, it may affect a part of the SCF receptor gene other than those altered by the dominant W mutations (Nocka et al., 1990), such as that encoding the extracellular domain, or a control sequence regulating expression. If the rs locus encodes a different receptor, two candidates are the PDGF receptor A and Flk-1 (Stephenson et al., 1991; Matthews et al., 1991; see Introduction). We do not know of evidence separating the rs and Ph loci, although rs is distinct from Rw because rs maps inside the W^19H deletion and Rw outside it (Geissler et al., 1988).

Reduced proliferation is also consistent with the observed phenotype of reduced numbers of hair-follicular melanocytes in si/si skin. No major increase in death rate was observed in culture, but this is not incompatible with the report of premature death of silver melanocytes in the hair cycle in vivo (Quevedo et al., 1981). This death may depend on periodic events related to the hair cycle, such as falling levels of growth- or survival-factors, or increasing melanocyte differentiation, which may not occur in culture. Mintz (1971) suggested - from the patterns of pigmentary mosaicism in chimaeric mice containing both wild-type and si/si tissue - that the silver mutation acted primarily on the follicles rather than the melanocytes. We however have observed a direct effect on cloned melanocytes. It is possible that the si gene product is a growth, survival or attachment factor synthesized both by melanocytes and by other skin cells, so that si melanocytes can function normally in wild-type skin (as implied by Mintz’s report), and wild-type melanocytes in mass culture can also sufficiently provide the factor, but isolated si melanocytes cannot. This would not however explain the interaction of the b mutation with silver. The product of the b locus is a presumptive melanosomal enzyme, tyrosinase-related protein 1 (TRP-1) (Jackson et al., 1990; Bennett et al., 1990), which alternatively suggests a melanosomal location for the si gene product. Moreover the product of pmel 17-1, the melanocyte protein that is a candidate for the si locus product, shows areas of sequence homology with TRP-1, the b locus product, and with tyrosinase (Kwon et al., 1991). This suggests that it may be another melanosomal enzyme, which could easily interact with TRP-1 in a competitive manner. It may seem surprising that a mutation in a pigmentary enzyme could inhibit cell growth, but for comparison there are two known blocus mutations that actually kill melanocytes as they differentiate in each hair cycle (Jackson et al., 1990).

We and others are now investigating the molecular defects present in rs/rs and si/si melanocytes, by various means including analyses of the integrity of parts of growth-signalling pathways that could be cell-type specific, and of the state of c-kit in rs/rs melanocytes and pmel 17-1 in si/si cells. The cell lines reported here should greatly assist the characterization of these two genes required for the development of a full complement of melanocytes in the skin.

This research was supported by the Wellcome Trust and the Cancer Research Campaign. We thank numerous colleagues for stimulating discussions and Byoung Kwon for communication of unpublished work.