Immunochemical demonstration of the clonal organization of chimaeric mouse epidermis

Monoclonal antibodies directed against H-2 antigens have been used as strain-specific markers for cellular genotype in various sectioned tissues of adult chimaeric mice (Ponder, Wilkinson & Wood, 1983a; Ponder et al. 1985). Here we report on the use of the marker system to visualize the chimaeric components in whole-mount preparations of mouse epidermis.

Chimaeric mice have been used to address a variety of questions about clonal histories in development and in adult animals (e.g. McLaren, 1976; Le Douarin & McLaren, 1984 for reviews; Schmidt, Garbutt, Wilkinson & Ponder, 1985a). Whole-mount preparations are better suited than tissue sections for this purpose because it is difficult to infer clone sizes in the intact tissue from the patterns of mosaic patches in tissue sections (Schmidt, Wilkinson & Ponder, 1985b; Gardner, 1984a).

The examination of mosaic patterns in sheets of chimaeric epidermis provides information about the normal pattern of development of this tissue and the clonal organization of epidermis in the adult mouse. Green, Durham, Mayer & Hoppe (1977) proposed, from observations of the pattern of mutant hair types in chimaeric coat, that the epidermis showed similar directional clonal expansion and coherent proliferation during development as had previously been suggested for melanocytes by Mintz (1967). In the adult, the epidermis of ear and body skin is thought to have a specific clonal architecture based on the ‘epidermal proliferative unit’ (Potten, 1974): the visible element of structure is found by precisely aligned columns of flattened cells in the spinous, granular and squamous layers. Beneath each column lie groups of approximately ten basal cells, 2-7 % of which are clonogenic (i.e. capable of generation as determined by radiobiological experiments). There is evidence that the single basal cell lying beneath the centre of the column is more slowly cycling than the cells lying beneath the edges. It is proposed that the central cell is the clonogenic stem cell, whose daughters move in the column, and, hence, emigrate in coordination with the interdigitating cells of adjacent columns to form the flattened cells of the column itself. As with other tissues (see above), such a highly geometrically clonal organization of the basal layer of the epidermis should be revealed in a chimaeric system.

Binomial mathematics were employed in a study of chimaeric epidermis (lannaccone, Gardner & Harris, 1978; lannaccone, 1980) which asked from how many cells experimentally induced skin tumours arise. In that analysis, mosaic patterns were not visualized by staining; instead, the composition of small tissue fragments was examined using electrophoretic polymorphisms. The variances of the proportions of the chimaeric components obtained were used to calculate an average clone size; a mean patch size was then estimated from the theoretical relationship between clone and patch size (West, 1975). Further calculations, also based on binomial sampling and incorporating the observed frequency of tumours with mixed genotypes, yielded estimates for a hypothetical target size (about eight cells), the number of cells that would give rise to a single tumour. The inherent difficulties with this type of statistical approach have been discussed in detail elsewhere (Mead, Schmidt & Ponder, 1987; Schmidt & Ponder, 1987; Wilkins, 1986). In order to exploit the potential of the chimaeric system for answering questions regarding tumorigenesis, clearly it will be crucial to know the actual distribution and sizes of patches rather than to rely on untested assumptions about them. The in situ staining method of whole-mount preparations of epidermis reported in the present paper will be of value to these applied studies.

CBA/CaLac (CBA; H-2^k) mice were bred in the animal house of the Institute of Cancer Research, C57BL/6JLac (B6; H-2^b) and AKR (H-2^k) mice were obtained from Olac Ltd (Bicester, UK). Eight chimaeras were obtained by aggregation of 4- to 8-cell embryos according to methods described by Mintz (1971): five B6↔CBA, two B6↔AKR and one B6↔RII1. The aggregated embryos were brought to term and reared by B6xDBA/2Lac F₁ - hybrid foster mothers.

Chimaeras were aged 6 weeks to 2 years when killed by cervical dislocation. The hair of the body (trunk: dorsal/ lateral/ventral) was removed using an electric shaver (Small Animal Clipper, A5-00, Oster Corp. Milwaukee, Wisconsin, USA) followed by a 20min application of Immac Cream (Anne French, London) and a final rinse in phosphate-buffered saline (PBS), pH 7·3. In order to loosen the epidermal and dermal layers, the sheets were transferred to 20mM-disodium ethylene (diamine) tetra acetate (EDTA) in Hanks buffered-salt solution (HBSS) adjusted to pH7·2, and left with gentle stirring for 19-22 h at 37°C. Following the EDTA treatment, the skin sheets were pinned out (stretched) in a dissecting dish, lined with black paraffin wax and containing PBS, and the remaining preparation was carried out under a dissecting microscope (SD2, Kyowa Optical) using a cold light source (KL, 1500, Schott, Wiesbaden). The epidermis (uppermost) was carefully separated from the dermis by holding down the relatively thick dermis with forceps while gently pushing back - and thus separating - the epidermis with the closed end of a second pair of forceps. The isolated epidermal sheets were transferred (basal surface uppermost) to a Petri dish lined with white paraffin wax, pinned down with fine entomological pins and fixed in 10mM-periodate-0·15M-lysine-1 % paraformaldehyde (PLP) pH7·4 for 2min.

The staining method was adapted from that developed for tissue sections (Ponder et al. 1983a). The monoclonal antibodies clone 11·4-1 (anti-H-2^k) and FT 6 ∝-9 (anti-H-2^b) were purified from ascites and conjugated directly to alkaline phosphatase and peroxidase as previously described (Ponder, Wilkinson, Westwood & Wood, 1983b). After fixation, the epidermal preparations were washed with PBS and immediately immersed in 0-1 % phenylhydrazine hydrochloride in PBS for 30 min to block endogenous peroxidase. After several washes in PBS and a final wash in PBS containing 0·5% bovine serum albumin (PBS-BSA), the preparations were incubated in 10 % mouse serum (from CBA mice) for 30 min before application of the first antibody. Best results were obtained (Ponder et al. 1983a) if the anti-H-2^k monoclonal antibody (11.4-1; conjugated directly with alkaline phosphatase) was applied before the H-2^b monoclonal antibody (FT 6 ∝ 9; conjugated directly with horseradish peroxidase). The epidermal sheets were incubated for 90 min at room temperature with each of the antibodies (diluted 1:50 to 1:200 in 10% mouse serum, pH7·6); preparations were washed in PBS-BSA between and after incubation. Peroxidase was demonstrated first, using 3’3’ diamino benzidine (DAB; Sigma) in 50mM-Tris pH 7·2 as substrate, yielding a brown reaction product. The alkaline phosphatase substrate was naphthol AS-BI sodium salt (Sigma) coupled to Brentamine fast red TR, in veronal acetate buffer pH9·2, yielding a red reaction product. Endogenous alkaline phosphatase was inhibited by addition of 1 mM-levamisole (Sigma) to the substrate mixture (Ponder & Wilkinson, 1981). Stained preparations were kept in 5 % formol saline, but the staining intensity became weaker after several weeks of storage.

Nonchimaeric B6 and CBA skin sheets were stained with both antibody conjugates simultaneously with a chimaeric sheet as controls.

Fig. 2A-G were taken with a Polaroid MP.4 Land Camera System holding a Contax (RTS) camera, using Kodak Ektachrome 50 ASA.

The resolution of the immunohistochemical staining was sufficient to allow us to identify unequivocally the chimaeric components at the level of a single epithelial cell (Fig. 1). Nonchimaeric, control sheets of epidermis stained homogenously for the respective H-2 haplotypes (Fig. 2G).

The relative proportions of the mosaic H-2 haplotypes were visually estimated to have varied between 0·3 and 0·7 (Fig. 2A,C,E). Thus, we did not obtain chimaeras in which the components were overall highly unbalanced (e.g. one component less than 0·05); such unbalanced animals are particularly suitable for quantitative clonal analysis of tissues in which the arrangements of clones show little obvious spatial pattern (Schmidt, Wilkinson & Ponder, 1985c, 1986b). In the epidermis, however, the clonal patterns observed were striking: transverse clonal stripes were observed in all specimens, although more pronounced in some than in others (Fig 2A,C,E). They extended from approximately mid-dorsal, laterally down both flanks; a dorsal midline was, however, not sharply defined (Fig. 2A). In epidermal sheets in which large dorsal/lateral areas were dominated by only one component, small transversely oriented, longitudinal patches of the other genotype were clearly delineated, probably representing single coherent clones (Fig. 2D). These patches were always smaller, often considerably so, than the obvious transverse bands in areas of epidermis with relatively more equal contributions of the two component genotypes (compare Fig. 2A,E). The edges of both stripes and coherent clones showed fine-grained mosaicism (Fig. 2B).

A very different clonal pattern was found for the belly region, the most conspicuous feature being a relatively marked patch boundary along the ventral midline (Fig. 2C,E). The length over which this demarcation line was visible varied depending on whether the adjoining regions were of dissimilar genotypes. In some animals it extended to 3 cm, whereas no line at all was visible in one specimen in which the entire belly region was dominated by the B6 component (Fig. 2A). The definition of the boundary was less pronounced when viewed at higher magnification (Fig. 2F), the irregularity extending over some 20 cell diameters. Other large areas of the belly were often dominated by either one of the two chimaeric components (Fig. 2A,C) and the striking patterns of clonal stripes observed in the dorsal region and flanks were absent.

Hair follicles lying within a patch of epithelium of a single parental type would be expected to be homogenously stained whether they arose from one or from more cells; but at the border between patches of differing parental type, mixed hair follicles would be expected if their origins were polyclonal. We observed that hair follicles at such boundaries were indeed often mixed.

There were irregularities of the patch boundaries and occasionally isolated patches of two or three cells in the skin of dorsal, flank and belly regions (e.g. Fig. 1) whose size and shape did not conform to that expected if the basal layer were composed of monoclonal epidermal proliferative units (EPU’s) as described by Potten (1974).

The marker system employed in the present study is suitable for positive identification of both parental components in whole-mount preparations of chimaeric mouse epidermis. Patches of dorsal/lateral but not ventral epidermis showed a directional preference of clonal expansion consistent in different animals. Hence, our study supports the suggestion made by Green et al. (1977) regarding the dorsal growth of epidermis (see Introduction), but it revealed a different clonal pattern for ventral epidermis. The clonal sizes observed in the epidermal sheet with relatively unbalanced components (Fig. 2D) suggest that most of the large conspicuous stripes in some other specimens (e.g. Fig. 2A) are due to aggregation of adjacent coherent clones of the same genotype into larger transverse patches (i.e. stripes) in a manner analogous to that described for melanocyte patterns (Silvers, 1979). Since the epidermis grows as a cell sheet, the stripy patterns may be interpreted as showing preferred orientations of mitoses, while in the coat, the migratory behaviour of the melanocyte population greatly contributes to the final patterns (Mintz, 1967).

The clonal pattern in the ventral epidermis was markedly different. There were no stripes anywhere from the midthoracic region to the groin, but a straight midventral clonal boundary was observed. These findings were consistent in chimaeras of each of the three strain combinations studied. The lack of obvious ventral striping in epidermis suggests that the growth of the epidermis is less-strongly directional here than it is dorsolaterally, perhaps because of the growth associated with the closure of the anterior abdominal wall. In contrast, striping in melanocyte chimaeras may extend to the midline in the upper abdominal region or over the belly (Fig. 3), implying that by the time of melanocyte migration, the direction of epidermal growth no longer leads to disruption of the dorsal/lateral striping pattern.

Both melanocytes and epidermal cells observe a well-defined midventral boundary. The straightness of the ventral midline in the epidermis appeared to be intermediate between the borders reflecting compartment boundaries in Xenopus (Jacobson & Klein, 1985) or Drosophila (Postlethwait, 1978), and the (presumably noncompartment) edges of clones in dorsal epidermis (Fig. 2B). The most probable explanation of the ventral midline boundary in epidermis is that it results from the physical isolation of the left and right halves of the ventral body wall during early embryogenesis before the ventral closure of the body wall at 9 dayspostcoitum. The relative straightness of the boundary in the adult indicates that cell mingling in this region does subsequently occur, but only over a region extending to about 20 cell diameters. The absence of a clear dorsal midline boundary in epidermis suggests that there is a greater degree of mingling in this region, following closure of the neural folds.

The present study is a further example of the very informative approach of in situ staining of whole-mount preparations using chimaeras. So far, tissues that have been studied by this method have revealed a considerable spectrum of clonal organization, with different characteristics for particular tissues (Gardner, 1984b, 1985; Ponder et al. 1985; Ponder, Schmidt & Wilkinson, 1986; Schmidt, Wilkinson & Ponder, 1984; Schmidt el al. 1985a,b,c, 1986a,b;,Schmidt & Ponder, 1987). The demonstration of pattern in chimaeric epidermis, i.e. clonal stripes, midventral line, large areas of predominantly one phenotype in tissues with otherwise relatively balanced components, is consistent with a non-random distribution of the two cellular genotypes (in the statistical sense, pattern is defined as a departure from a random (Poisson) distribution of the entities under study (see Schmidt et al. 1985c)). The nonrandom mosaicism found is in contrast to assumptions made and results obtained by lannaccone et al. (1978; see Introduction). In that study epidermal fragments (0·5-1 mm²) were analysed electrophoretically and 92 % of a total sample of 495 contained contributions of both chimaeric components, albeit in varying proportions. This would indeed imply a fine-grained mosaicism and a possible random distribution of small patches as was suggested by lannaccone et al. (1978). Although in some regions of the epidermis of our chimaeras (e.g. Fig. 2B,F to left of midline) a fine mosaicism was present, in others (Fig. 2C) large areas were uniformly of one parental type or the other. The discrepancy between lannaccone et al.’s results and ours remains unexplained, although one possibility is the different strain combinations used to make the chimaeras (McLaren, 1976).

The chimaeric system has considerable potential for testing the clonality of anatomical structures. For example, whereas, in the adult, epithelial ceils of individual intestinal crypts are always of the same parental type (Ponder et al. 1985), hair follicles are mixed. We can conclude that the epithelium of the former, but not the latter, is derived from a single progenitor cell. The chimaeric system can also demonstrate the migratory pathway taken by cells from proliferative through functional anatomical zones (Schmidt et al. 1985b). Cells originating in the epidermal proliferative, basal layer are believed to migrate vertically in precisely defined columns of cells, several columns giving rise to well-determined geometric patterns (EPUs; see Introduction). If EPUs were monoclonal the mosaicism of chimaeric epidermis would be expected to reveal also a well-determined geometric pattern at patch boundaries and in areas where one of the parental components is in the minority. We did not, however, observe such regular patterns and conclude that our data indicate a polyclonal composition for EPUs. A polyclonal composition would not be consistent with the view of the centrally located basal cell being the sole progenitor of all cells in a single EPU.

We thank Mrs M. M. Wilkinson for her valuable assistance in a pilot study. This project was carried out during the tenure by B.A.J.P. of a Career Development Award from the Cancer Research Campaign and was supported by grants from the Medical Research Council, the Cancer Research Campaign, the Deutsche Forschungsgemein-schaft and the Bundesministerium für Forschung und Technologie.