Changing patterns of cytokeratins and vimentin in the early chick embryo

The intermediate filaments are a family of cell-type-specific proteins that appear to be expressed by all cells so far described (for reviews see Anderton, 1981; Lazarides, 1980, 1982; Osborn & Weber, 1982; Traub, 1985). During the first stages of embryogenesis only vimentin (mesenchyme-specific) and cytokeratins (epithelium-specific) have been observed. These intermediate filaments have been identified even at the oocyte stage in Xenopus (Gall et al. 1983; Franz et al. 1983; Godsave et al. 1984a,b; Klymkowsky et al. 1987). At later stages of Xenopus development, some of the first genes to be expressed that are significantly different from the maternal genome are those encoding for the cytokeratins (Jonas et al. 1985; Winkles et al. 1985; Sargent et al. 1986). In the mouse, cytokeratins and their mRNAs have been detected in the oocyte and preblastocyst stages (Lehtonen et al. 1983; Duprey et al. 1985; Lehtonen, 1985) and in stages prior to primitive streak formation (Jackson et al. 1980, 1981; Paulin et al. 1980; Brûlet et al. 1980; Kemler et al. 1981; Oshima et al. 1983; Duprey et al. 1985; Chisholm & Houliston, 1987), but vimentin does not appear until the primary mesenchymal cells have formed (Franke et al. 1982a). Similarly, in quail embryos (Erickson et al. 1987) and chick blastoderm cell cultures (Biehl et al. 1985) only cytokeratins and vimentin are expressed in the initial stages. Differentiation of muscle, neurones and glial cells, at later stages of development, is associated with the first appearance of the other intermediate filaments; these being desmin, neurofilament protein and glial fibrillary acidic protein, respectively (Jackson et al. 1981; Raju et al. 1981; Tapscott et al. 1981; Bignami et al. 1982; Franke et al. 19826; Tokuyasu et al. 1984; Bovolenta et al. 1984; Erickson et al: 1987). The early appearance of the cytokeratins in embryogenesis has led to the proposal that the formation of epidermis/ epithelium is one of the primary events in development, although why this is so is unclear (Sargent et al. 1986).

Despite a relatively comprehensive understanding of the biochemistry of the intermediate filaments, their precise function and the significance of their cell-type specificity is still equivocal. The cytokeratins might have a special function because of their association with desmosomes (Franke et al. 1984; Cowin et al. 1985; Jones & Goldman, 1985) in that they maintain the integrity of epithelial sheets. Experiments that have attempted to elucidate intermediate filament function by microinjection of intermediate filament-specific antibodies were notable by their failure to disturb normal cell behaviour (Eckert et al. 1981; Gawlitta et al. 1981; Klymkowsky, 1981; Lane & Klymkowsky, 1982; Lin & Feramisco, 1981). This has resulted in a recent challenging theory which proposes that the intermediate filaments might act in the regulation of gene expression (Traub, 1985). This idea has arisen from experimental data which demonstrate that intermediate filament subunit proteins have a high affinity for nucleic acids (Nelson & Traub, 1981; Traub & Nelson, 1982).

As histogenic markers, intermediate filaments can provide useful information in determining cell differentiation events during embryogenesis. The chick embryo is particularly valuable in this respect because our knowledge of cell and tissue movements in its early development is more extensive than in any mammalian embryo. Studies on the appearance of intermediate filaments during development might also provide clues towards a functional rôle. Hence, this study describes the distribution of cytokeratins and vimentin in the early chick embryo using immunohistochemical methodology.

Hens’ eggs (Ross Brown, Andover, UK) were incubated in a rocking incubator at 38°C in humid conditions. The embryos were then removed from the yolk and vitelline membranes at the appropriate times to give a range of embryos between stages 3 and 12 according to Hamburger & Hamilton (1951). These embryos were placed in Pannett & Compton saline (Pannett & Compton, 1924) and excess yolk removed. They were then fixed in either absolute ethanol for 15 – 30 min or in a solution containing 3 – 7% paraformaldehyde, 50% ethanol and 4 % glacial acetic acid for 45 min. Ethanol-fixed embryos were rehydrated through a graded series of ethanols (90 %, 70 %, 50 %) to PBS, followed by washes with 5 % w/v sucrose in PBS for 2h and finally 15 % sucrose-PBS. Those fixed in paraformaldehyde were washed in PBS and taken to 15 % sucrose-PBS as above. Tissues were stored at 4°C in this solution to which sodium azide had been added (final concentration 0 · 01%) either overnight or up to 2 weeks. Subsequently, the fixed embryos were infiltrated with 7 – 5% gelatin (300 Bloom; Sigma Chemical Co., Poole, UK) in 15 % sucrose-PBS for 2 h at 37°C, allowed to set and then frozen in isopentane previously cooled with liquid nitrogen. Transverse serial sections were cut at 8 – 10 pm in a Bright cryostat set at —25°C and then mounted on gelatinized slides and stored at 4°C until use.

Tissue sections were rinsed in distilled water for 10 min and then either single- or double-labelled for 15 min to 1 h at 37°C with anti-cytokeratin and anti-vimentin antibodies. The antibodies were gifts and are detailed in Table 1. In doublelabelling experiments, either one of the anti-cytokeratin antibodies was used with the anti-vimentin antibody. Single labelled sections were subsequently overlaid with either goat anti-mouse IgG or goat anti-rabbit IgG antibody (as appropriate for the primary antibody) conjugated with fluorescein isothiocyanate (Sigma Chemical Co.) for 30 min at 37°C in the dark. Double-labelled sections were incubated in a cocktail containing goat anti-mouse IgG and goat anti-rabbit IgG antibodies conjugated with rhodamine and fluorescein fluorochromes (Sigma), respectively. These second layer antibodies had previously been passed down Sepharose columns containing bound rabbit or mouse IgGs (as appropriate) to remove cross-reacting antibodies. Finally, the sections were washed in distilled water (three changes) and mounted in an aqueous mounting medium (Citifluor, City University, London, UK). They were examined under a Zeiss photomicroscope using epifluorescence. Photographs were taken on Kodak Tri-X Pan 400 and Ektachrome 400 rated at 800 ISO.

In all immunohistochemical procedures, a group of sections was included in which either the first, second or both antibodies were omitted from the protocol. Specific fluorescence was not observed in these sections.

Intermediate-filament-enriched samples were prepared from embryos at various stages of development; these being 30 h, 3 days and 8 days of incubation. For each sample, three dozen embryos were dissected from the yolk and vitelline membranes in calcium- and magnesium-free Tyrode’s solution containing 5ITIm-EDTA and 01 mm-phenylmethylsulphonyl fluoride (PMSF). They were then homogenized by hand with a Douncer (at least 30 strokes) and spun at 11000g for 10min. The resultant pellet was resuspended in 0 · 05m-Tris buffer (pH8 · 0) containing 1% Nonidet P40 (NP40), 0 · 15m-NaCl and 5mm-EDTA. The sample was respun at 11000g for 10 min and the pellet dissolved in gel loading buffer (0 · 065 m-Tris (pH6 · 8) containing 2% sodium dodecyl sulphate, 5% ft-mercapto-ethanol and 10 % glycerol) and heated to 100 °C for 2 – 3 min.

Fractions were also prepared from the skin of prehatch chicks (eggs incubated for 20 days). The skin was cleaned of feathers, fat and muscle in PBS containing 0 · 1 mm-PMSF, cut into small pieces and homogenized in a Waring blender. The skin homogenate was then spun at 15000 g for 10 min and the pellet resuspended in PBS containing 1 % NP40 and then sonicated, all at 4°C. The suspension was respun and the pellet suspended in gel loading buffer and heated to 100 °C for 2 – 3 min. These samples were stored at —20°C until use.

The skin and embryo samples were run on standard SDS-PAGE (10% acrylamide), transferred to nitrocellulose paper and labelled overnight at 4 °C with the anti-vimentin and anti-cytokeratin antibodies. The labelled proteins were subsequently visualized by incubation with ¹²⁵I-donkey antirabbit IgG (Sigma) and exposure to X-ray film (Fujii). A rabbit anti-mouse IgG sandwich was included for those blots labelled with the anti-cytokeratin antibodies.

Nine stage-12 embryos (48 h incubation), dissected free of yolk and vitelline membranes, were each placed in a well of a Multiwell tissue culture plate (Falcon plastics, type 3008) containing 100 μ l of methionine-free Dulbecco’s modified Eagles medium (DMEM). They were incubated in this medium for 1 h at 37°C (5 % CO2) to deplete endogenous methionine pools, furthermore, most of the area opaca was trimmed from these embryos to reduce the methionine input from yolk granules. The medium was pipetted off and replaced with 100 μl of labelling medium containing a 1 in 20 dilution of L-[³⁵S]methionine (Amersham; SJ 1515,1000Ci m-mol^-1) in methionine-free DMEM. This gave a final concentration of 0’75mCiml^-1. The embryos were then incubated for a further 4h at 37 °C.

Preparation of embryo lysates and immunoprecipitations followed the method of Oshima (1981). The medium was withdrawn and the embryos washed three times in cold PBS followed by addition of 1 ml per embryo of SDS harvest buffer (01% SDS, 10mm-Tris-HCl (pH7 · 4), 3 mw-MgCh, 01 mm-CaCl₂) containing 0 · 5 mm-PMSF, 1 mm-N-ethymaleimide and 1/100 vol. of aprotinin solution (Sigma), all at 4 °C. The embryos were pooled and dounced in the harvest buffer followed by the addition of 45 μl DNase 1 (l–2 mg ml^-1; Sigma). Further SDS was added to 0 · 5 % and EDTAadded to 1 HIM final concentrations. The embryo lysate was then heated at 100°C for 2 min and, after cooling, Nonidet P-40 added to 1 %. Aliquots (4 μl) of the lysate were taken for determining TCA-precipitable counts and the remainder stored frozen at —20°C. Counts ranged from 568 × 10⁴ to 5 · 73 × 10⁴ dpm (mean = 5· 7 × 10⁴).

Lysates were incubated with the antibodies for 2h at 4 °C followed by a rabbit anti-mouse IgG sandwich for 1 h. Subsequently, formalin-fixed Staphylococcus aureus Cowan strain I bacteria were added and incubated for 30min. The bacteria were recovered by centrifugation and washed twice in 01% SDS, 1% NP-40, 10mm-Tris-HC), pH 7 · 4, 5mm-EDTA, twice in 0-5m-NaCl, 50mm-Tris-HCl, pH 7 · 4, 5mm-EDTA, 0 · 05% NP-40 and resuspended in 0 · 15m-NaCl, 50mm-Tris-HCl, pH 7 · 4, 5mm-EDTA, 0 · 05% NP-40. The radioactive proteins and immunoglobulins were solubilized in gel loading buffer, heated to 100°C for 2 min and separated on polyacrylamide gels. The gels were processed for fluorography with 2,5-diphenyloxazole-dimethyl sulphoxide and exposed to Kodak X-ray film at — 70°C.

The patterns of immunofluorescence following treatment with LP1K, LP3K and LE65 antibodies were identical in all stages examined and are described together. Fixation in absolute ethanol gave optimal immunofluorescence for the anti-cytokeratin antibodies but morphological preservation was poor; the use of other fixatives drastically reduced or completely abolished antigenicity. The anti-vimentin antibody was less susceptible to the fixative used, but those embryos fixed in paraformaldehyde gave superior immunofluorescence and morphological preservation.

The specificity of the anti-intermediate filament antibodies was analysed by immunoblotting and immunoprecipitation. LP3K and LE65 antibodies did not react with blotted proteins or precipitate protein from radiolabelled embryos (data not shown); LP1K, however, was found to recognize a single band of an apparent molecular weight of 56 000 in intermediate filament-enriched material prepared from embryos incubated for 30h, 3 days (not shown) and 8 days (Fig. 1). This protein band was not resolved in skin fractions (Fig. 1).

The anti-vimentin antibody recognizes a major band of apparent molecular weight 58000 in immunoblots of embryonic material (30 h, 3 days (not shown) and 8 days) and chick skin (Fig. 1). Immunoprecipitation data confirmed this finding in stage-12 embryos (Fig. 1).

The primitive streak is the first readily observable structure to form at around 6h of incubation and by about 19 h extends two thirds of the way across the area pellucida along the craniocaudal axis (Hamilton, 1952) (Fig. 2A). At this stage the embryo consists of three germ layers; the ectoderm, mesoderm and endoderm (Fig. 2B).

LP1K/LP3K/LE65 antibody binding at these stages, was restricted mostly to the ectoderm cells at the perimeter of the embryonic disc overlying the area opaca and edge of the area pellucida (marginal zone) (Fig. 3). In these early stages, the apical regions of the cells were preferentially labelled (Fig. 3). A few cells in the mesoderm and endoderm were also labelled (Figs 3, 4); those in the endoderm were restricted to peripheral regions (Fig. 3).

Anti-vimentin antibody labelling of the ectoderm was observed in central regions of the embryonic disc, including the primitive streak (Fig. 4), but only occasional cells were labelled in the peripheral ectoderm overlying the area opaca and marginal zone (Fig. 3). The mesoderm and endoderm showed moderate fluorescence (Figs 3,4).Double-labelling experiments revealed that anti-cytokeratin and anti-vimentin antibody distributions were largely mutually exclusive except for occasional cells in the ectoderm and endoderm (Figs 3, 4).

The embryo at this stage possesses a head process, Hensen’s node and a definitive primitive streak, but is still composed three germ layers (Fig. 5). The neural plate develops following the craniocaudal regression of Hensen’s node and subsequent notochord formation. It can be identified as a layer of columnar cells in the ectoderm which is henceforth distinguished as either neural or non-neural (Fig. 5B). The head fold begins to form at about stage 6 and the first pair of somites develop at stage 7.

At these stages, the distribution of immunofluorescent cells in the ectoderm and endoderm is more extensive than in younger embryos. Immunoreactivity persists in peripheral regions but it has also progressed centripetally. In rostral regions, the non-neural ectoderm is labelled entirely, but the neural ectoderm (neural plate) is unreactive apart from the cells at its lateral margin (Figs 6, 7). Similarly, fluorescent cells in the endoderm are present both peripherally and centripetally, but only a few cells are labelled in the midline (Figs 6, 7). In caudal regions, the ectoderm and endoderm exhibit immunoreactivity peripherally and centripetally but not in the midline (Fig. 8).

Anti-vimentin antibody labelling of the neural and non-neural ectoderm exhibited a reversed staining pattern to that for the anti-cytokeratin antibodies. Hence, the neural plate is immunostained throughout (Fig. 6), but the non-neural ectoderm is generally unreactive apart from a few isolated cells (Fig. 7). In caudal regions, binding of the anti-vimentin antibody to the ectoderm, at these stages, becomes more restricted to centripetal regions of the embryonic disc and only limited fluorescence is seen (Fig. 8). The endoderm and mesoderm however, are labelled in all areas of the blastoderm (Figs 6, 7, 8).

Double-labelled sections showed areas of fluorescent overlap in the ectoderm, the endoderm and at the lateral margins of the neural plate (Figs 7, 8), but, as in earlier stages, the immunostaining patterns showed an overall mutual exclusiveness.

During these stages Hensen’s node continues to regress and to form notochord (Fig. 9). The head fold becomes more pronounced and the heart primordia develop to fuse eventually in the midline, forming the heart (Fig. 10). The folding of the endoderm during heart development results in the formation of the foregut (Fig. 10B,C). The somites continue to form in a craniocaudal sequence from the segmental plates situated either side of the neural tube (Fig. 9).

Caudally, the primitive streak persists and the pattern of immunoreactivity in the ectoderm and endoderm is identical to that seen at earlier stages. In more cranial sections, however, both the ectoderm and endoderm are fluorescent in all regions (Figs 11A, 12A, 13, 14). The fluorescent ectoderm can be distinguished clearly from the unlabelled neural tube (cf. unlabelled neural plate, Fig. 6), but where it has not closed dorsally, some cells (neural crest/tube?) lying on the medial aspect were also labelled (Fig. 12A). Once the tube has closed, as seen in more cranial sections and at later stages, only the ectoderm is fluorescent (see Fig. 11A). In the region of the heart folds, the endoderm and associated splanchnic mesoderm are strongly labelled (Fig. 13). The structures that are derived from these layers (the foregut and heart, respectively) also exhibit immunoreactivity (Fig. 14). The precardiac (splanchnic) mesoderm develops into the epimyocardium and endocardium, but only the former is immunoreactive (Fig. 14). The somatic mesoderm is also fluorescently labelled (Fig. 13).

The notochord, neural tube, somites and segmental plate are unreactive (Figs 11A, 12A, 14), but the somatic and splanchnic components of the lateral plate (Fig. 11A) and the extraembryonic membranes (not shown) are strongly fluorescent.

Anti-vimentin antibody labelling at these stages is more extensive and is seen in most tissues that have differentiated, such as the somites, segmental plate, lateral plate and neural tube (Figs 11B, 12B). In these epithelial tissues, the fluorescence showed a marked baso-lateral distribution. The ectoderm is weakly labelled but the endoderm shows strong fluorescence in all areas of the embryonic disc (Figs 11B, 12B, 13, 14). The splanchnic and somatic mesoderm in the heart fold region also exhibit fluorescence (Fig. 13) and as the heart develops both the epimyocardium and endocardium are labelled (Fig. 14). In the lateral plate, the splanchnic mesoderm is more strongly labelled than the somatic component (Fig. 11B).

Double-labelled sections revealed large regions of fluorescent overlap in the splanchnic and somatic mesoderm, epimyocardium, foregut, endoderm (Figs 13,14) and lateral plate (not shown, but compare Fig. 11A and 1113).

The main findings of this paper demonstrate that cytokeratins and vimentin intermediate filaments are expressed at early stages of chick embryogenesis. The anti-cytokeratin antibodies utilized in this study recognize the human cytokeratins designated CK7, CK8 and CK18 (Moll et al. 1982). These polypeptides are charac-teristically expressed in embryonic and simple epithelia in which the CK8 and CK18 proteins interact specifically to form 8 nm keratin filaments (Fuchs et al. 1984); CK7 can also interact with CK18 (Sun et al. 1984). The codistribution of these cytokeratins in chick embryonic epithelia concurs with these findings. The results indicate that the LP1K antibody recognizes a filamentous, intracellular protein in embryonic tissues of apparent molecular weight 56000 which is not expressed in the complex epithelium of the chick skin. Such data provide persuasive evidence that the LP1K, LP3K and LE65 antibodies specifically recognize a cytokeratin filament in chick embryos.

The anti-vimentin antibody recognizes a 58000 molecular weight protein in chick tissues as shown by immunoblotting and immunoprecipitation (Fig. 1); this agrees with previous findings using this antibody (Jacobs et al. 1982).

The initial patterns of expression of cytokeratins and vimentin in the primitive streak stages of the chick embryo can be related to the cell migrations that occur at this time (see Bellairs, 1986). There appears to be only limited coexpression of vimentin and cytokeratins. The cytokeratins are restricted, at first, to the ectoderm overlying the area opaca and marginal zone and are associated with cells destined to remain in the ectoderm, whereas vimentin is localized to centripetal regions of the ectoderm and is associated with cells destined to ingress through the streak to form primary mesoderm and definitive endoderm. The subsequent expression of vimentin in mesodermal and endodermal tissues may therefore be attributed to these migratory cells.

The distribution of cytokeratins at the primitive streak stage correlates with that of desmosomes which are restricted to the peripheral ectoderm (area opaca) and marginal zone of the area pellucida (Overton, 1962; Bellairs, 1986; Bellairs et al. 1978). It is likely that the desmosomes and cytokeratin filaments are closely associated in this ectoderm for the following reasons: (a) the cytokeratins and desmosomes are apically distributed, (b) the appearance of desmosomal proteins (desmoplakins) and cytokeratin filaments are correlated during development (Jackson et al. 1980) and (c) the desmosome-tonofilament (cytokeratin) complex is characteristic of epithelia (Franke et al. 1978, 1984; Jones & Goldman, 1985). The function of this complex is to maintain the integrity of the epithelial sheet and must be a vital requirement for the chick blastoderm undergoing centrifugal tension during the primitive streak stage (Bellairs et al. 1967).

The initial absence of cytokeratins in the chick endoderm is considered unusual because this epithelium, along with the ectoderm, is one of the first cell layers to be formed. However, the endoderm may express other cytokeratin filaments, characteristic of simple epithelia (e.g. CK19 in humans; Sun et al. 1984) which are not recognized by the antibodies used in the present study. The endoderm is initially squamous rather than cuboidal (as in the ectoderm) and only later is a cuboidal morphology assumed, by which time immunoreactivity is apparent. Hence, epithelial cell morphology may reflect the type of intermediate filament expressed (see Connell & Rheinwald, 1983; Ben Ze’ev, 1984) and also the cytokeratin composition (see Quinlan et al. 1985).

The appearance of vimentin in the endoderm and mesoderm may be related to the motile behaviour of these cells as they detach from the ectoderm during ingression. Other studies have also associated vimentin expression with reduced cell-cell contact, cell spreading and growth (Connell & Rheinwald, 1983; Ben Ze’ev, 1984). These authors have demonstrated that epithelial cells switch their normal cytokeratin synthetic pattern to a predominantly vimentin profile in low cell densities and vice versa at high cell densities.

In the mouse embryo, the first appearance of vimentin is seen in mesenchymal cells only (Franke et al. 1982a). Cytokeratins are expressed across the entire ectoderm and the desmosome-cytokeratin complex is lost as cells leave the primitive streak. Variations in ingression events between mammals and birds (see Poelmann, 1981) might account for these differences.

Differentiation of the ectoderm into neural and non-neural components shows a marked correlation with the appearance of vimentin and cytokeratins, respectively. There are only small regions of coexpression in the lateral margins of the neural plate. The absence of cytokeratins in presumptive neural tissue is consistent with other studies (Tapscott et al. 1981; Erickson et al. 1987) but is perhaps surprising in an epithelium that undergoes severe shape changes and hence would require structural unity. This may however, be provided by the basolaterally distributed vimentin filaments associating with hemidesmosomes (see Kartenbeck et al. 1984).

Closure of the neural tube is associated with the appearance of cytokeratins in the neural crest and might be a preparatory event for ectodermal sealing by formation of desmosome-tonofilament complexes.

Subsequent expression of cytokeratins in regions other than the ectoderm is complementary to vimentin expression, for example, in the endoderm, lateral plate and precardiac mesoderm (later to become the epimyocardium). In these regions, the appearance of cytokeratins might be indicative of an ‘epithelialization’ event. In the heart folds, for example, both the mesoderm and endoderm experience tensions during the folding process and hence there is a requirement for an integral sheet to withstand these forces.

The early differentiation events in the chick embryo are not associated with transitions of expression from one intermediate filament type to another, whereas at later stages of development, for example, there is an apparent replacement of vimentin by desmin during myogenesis (Bennett et al. 1979; Holtzer et al. 1981) or by neurofilament protein during neurogenesis (Tapscott et al. 1981). Vimentin is not replaced by cytokeratins during formation of the foregut and heart for example, and transitions of intermediate filament expression appear to be related therefore to terminal differentiations of functional adult tissues such as nerve and muscle. Coexpression of cytokeratins and vimentin may be an important feature of the relatively undifferentiated tissues of the embryo and appears to be a widespread trait that has not been fully recognized (see Lane et al. 1983).

As cell-type-specific markers, the distribution of intermediate filaments should reflect histogenic lineage. The expression of cytokeratin and vimentin filaments during chick embryogenesis, however, appears to be related more to functional and behavioural requirements rather than to germ layer derivation (see Erickson et al. 1987).

This work was funded by Action Research for the Crippled Child (Grant No. A/8/1526) and The British Heart Foundation (Grant No. 86/88). I am grateful to Drs B. Lane and P. Hollenbeck for providing antibodies, to Ros Cleevely and Liz Harfst for expert technical assistance and to Professor Ruth Bellairs for reading the manuscript.