Mouse neural crest cells secrete both urokinase-type and tissue-type plasminogen activators in vitro

The two serine proteases, urokinase-type (uPA) and tissue-type (tPA) plasminogen activators, are involved in extracellular proteolysis (Liotta et al. 1981; Reich et al. 1988). Both enzymes are thought to play a role at the time of cell migration or tissue remodeling and in thrombolysis (for review see Dano et al. 1985; Reboud-Ravaux, 1985; Saksela, 1985). They convert the only known substrate, the ubiquitous zymogen plasminogen, into the trypsin-like protease plasmin. Plasminogen activators (PAs) are secreted by many cells in vivo, as well as by cells of primary cultures or established lines. In particular, migratory and invasive tumor cells produce PAs (Carpén et al. 1986; Hoal-Van Helden et al. 1986), and a significant correlation was found between uPA mRNA content and metastasis (Sappino et al. 1987) .

In mouse, the presence of PA has been demonstrated in several adult tissues (Danglot et al. 1986; Larsson et al. 1984; Huarte et al. 1985; 1987). In embryos, PA activity has been detected in trophoblast giant cells as well as in other cells (Bode & Dziadek, 1979; Marotti et al. 1982). At day E8-5 (vaginal plug at day E0), the neural crest cells (NCC) begin to migrate extensively from the area of neural tube closure to their site of differentiation. Several physical barriers need to be disrupted for migration to take place, such as the basal lamina that underlies the neural tissue, and the network of fibronectin and other extracellular matrix (ECM) proteins. One might therefore suspect that a protease activity is necessary for this process of migration, for which PAs are potential candidate enzymes. In accord with this hypothesis, quail embryo NCC have been shown to produce PA in vitro (Valinsky & Le Douarin, 1985). Moreover, the interaction in vitro between tPA and fibronectin or laminin (Salonen et al. 1984; 1985) suggests a direct relationship between the presence of PA and the extracellular proteolysis during cell migration; in this context it is important to remember that fibronectin was demonstrated to be an important ECM glycoprotein for attachment and migration of NCC (Thiery & Newgreen, 1980; Thiery & Duband, 1986; Dufour et al. 1988a, b).

Here, we show that cultures of NCC from mouse embryos produce both tPA and uPA. The highest level of uPA secretion was found in the most caudal segment of the developing neural system.

Embryos at day E8·5 of gestation (stage 13, 6-10 somite pairs according to Hogan et al. 1986) were collected from Swiss females and dissected in MEM (Gibco) essentially as described by Jaenisch (1985). Briefly, the posterior trunk regions, including the last four somites, were cut and incubated in 1% trypsin (Sigma 1:250) in Tris-phosphate-buffered saline (Hogan et al. 1986) for 10 min at 4°C. The enzyme digestion was stopped in serum-supplemented MEM. The pieces of embryos were then pipetted up and down with a taped Pasteur pipet to separate the neural tube pieces from adherent tissues. After three rinses, they were cultured in microdrops (50 μ1 of culture medium) under light paraffin oil (Fisher 0-119) in 35mm culture dishes (Nunc). When not otherwise stated, the NCC were allowed to migrate out of the truncal neural tube expiants for 48 h in 2% Ultroser G (Gibco) in MEM.

Jaenisch (1985) has shown that isolated NCC cultured in DMEM supplemented with 5 % fetal calf serum (FCS) and 5 % horse serum were able to participate in normal development and differentiate into melanocytes upon microinjection in utero into E9 embryos. To attest the identity and potentiality for differentiation of the NCC in our culture conditions, Jaenisch’s experiment was repeated. Briefly, truncal neural tubes from pigmented inbred CBA embryos were isolated and cultured for 48 h in MEM supplemented with 2 % Ultroser G. Emigrated NCC were then collected by trypsinization and microinjected in utero into unpigmented inbred Balb/c embryos at E8·75. The injected embryos were allowed to develop to adulthood. They were then checked for the presence of pigmentation. CBA-derived NCC, cultured for 7-10 days in 10% FCS-supplemented MEM, were also analyzed for the presence of catecholamine-positive cells by formaldehyde fluorescence-induced (FIF) procedure. The cultures were washed with PBS, quenched in isopentane cooled with liquid nitrogen, lyophilized for Ih, and exposed to formaldehyde gas at 80°C for 1 h (Le Douarin, 1982; Ito & Takeuchi, 1984). The presence of neurofilament-positive cells was also investigated by immunofluorescence with indirectly labelled antineurofilament monoclonal antibodies from Dakopatts (Baroffio et al. 1988). The test was performed according to the manufacturer’s protocol.

After 48 h in culture, the neural tubes were detached from the dish with a platinum needle and discarded. The cells that had emigrated were rinsed twice with DMEM and cultured for 12 h in DMEM. The serum-free supernatant was collected and stored at —20°C and, when convenient, studied by electrophoresis in 10% SDS-PAGE at 8 mA. Serum-free media of PYS (Marotti et al. 1982) and MSV-3T3 (Belin et al. 1984) cells (generously provided by Dr J.-D. Vassalli, CMU, Geneva) were used as references for mouse tPA and uPA, respectively. The zymography was then performed as described by Vassalli et al. (1984), with slight modifications. Briefly, the acrylamide gel was rinsed twice in 2·5 % Triton X-100 in H₂O, twice in 50mm-Tris-HCl pH7·5 for 10min each and laid over an agar gel containing 1·3% commercial nonfat dry milk, 50mm-Tris-HCl pH7·5, 40 μg ml^-1 plasminogen and 0·8% agar. The zymograms were allowed to develop at 37°C for 24-72h. Plasminogen was purified from human plasma (Deutsch & Mertz, 1970). Standard curves were established with human urokinase (Sigma U-1627) by zymography. The lysis areas were measured and reported in Plough units. To express activity per cell, the cultures were photographed and the cells counted.

To detect PA activity of the cells directly in culture, a substrate overlay procedure was used. The agar gel was laid at 42°C on either primary or secondary NCC cultures. After 3 to 6 h of incubation at 37 °C, the overlaid cultures were examined for lysis areas.

The culture supernatant (20 μ1) was mixed with either 1 μ1 of anti-human tPA IgG (kindly provided by W.-D. Schleuning), or 1 μ1 of anti-mouse IgG (kindly provided by J.-D. Vassalli), or 1 μ1 of irrelevant IgG. The concentration of antibodies was 1 mg ml^-1. After 2 h at 4°C, 10 μ1 of a one-tenth suspension of fixed S. aureus (Sigma) was added to the samples. After 30 min of incubation at room temperature, the samples were centrifuged at 1000g for 5 min. The supernatants were mixed with 2x nonreducing buffer (lx = 50mm-Tris-HCl pH6·8, 2·5% SDS, 30% glycerin) and the pellets were washed with NET-TS solution (0·2m-NaCl, 10mm-EDTA, 50mm-Tris-HCl pH8·l, 1 % Triton X-100, 0·2 % SDS) and eluted in 20μ1 lx nonreducing buffer. The supernatants and eluates were run on SDS-PAGE and the gel was analyzed by zymography (Huarte et al. 1985).

The presence of NCC in our cultures was confirmed by the creation of chimeras, and by histochemical and immunohistochemical identification of markers for specific differentiation pathways. Two pregnant female mice were anesthetized and laparotomies were performed to expose their uteri. From the 11 Balb/c embryos, each microinjected with 50-100 CBA truncal NCC, 7 pups developed to term and 1 showed pigmentation derived from injected cells at adulthood. After 7-10 days of culture in 10 % FCS-supplemented MEM, cells derived from CBA-truncal neural tubes were found to develop neuron- and ganglion-like structures. In addition, catecholamine-positive cells were observed after FIF reaction and neurofilament-positive cells by immunofluorescence with monoclonal anti-neurofilament. These observations confirm the presence, in our cultures, of NCC with potentiality to differentiate into melanocytes and neuronal cells.

Altogether, 398 truncal neural tubes were isolated from E8-5 mouse embryos and cultured. In some experiments, pools of 10 tubes were made to minimize individual variations. For practical reasons (larger number of embryos, better synchronization of development, lower cost), outbred Swiss mice were used for the identification of PA activities in vitro.

After 48 h of incubation, a population of stellate cells well separated from each other was observed around the truncal neural tube expiant (Fig. 1). A second wave of confluent cells, extending from the explant, was also observed. The stellate cell population, considered as NCC, was tested for PA production (see below).

Cultured truncal neural tubes produce 200-500 NCC with low PA secretion. The microdrop culture system kept the enzymes secreted during 12 h in a small enough volume to be analyzed by zymography. The activities thus detected were weak, but clearly above the limit of sensitivity of the zymograms (about 01 milliPlough unit, in our hands).

When 48 h-old NCC were cultured for 12 h in serumfree medium, the presence of both tPA (72x10³M_r) and uPA (48x10³M_r) was detected in supernatant by zymography (Fig. 2, lane 1 and 3). The identity of both enzymes was confirmed by comparison with the mouse PAs references (Fig. 2, lane 10) and by immunoprecipitation with the specific antibodies (Fig. 2, lane 2 and 4-9). An additional lysis area at the position corre-sponding to an apparent 11Ox10³M_r was also observed (Fig. 2, lane 1). This material was immunoprecipitable by antibodies raised against human tPA (Fig. 2, lane 2). Similar high-relative-molecular-mass forms have been previously observed and demonstrated to be inhibitor-PA complexes (Danglot et al. 1986). The cells were also rinsed, scraped, lysed in 0·2% Triton X-100 and analyzed by zymography. The cell extract showed more tPA activity than the supernatant of the corresponding culture; no inhibitor-tPA complex was detected (data not shown).

The substrate overlay procedure, applied directly on the primary cultures, did not allow the identification of the cells that expressed PA. In fact, the PA secreted by the neural tube and attached to the bottom of the tissue culture dish developed a high lysis activity that over-whelmed the expected activity of the NCC. In order to avoid this problem, 48 h NCC were collected by trypsinization, subcultured for 3 h, and overlaid by a substrate-containing agar gel. The cells that secrete PA were recognized by a small lysis area surrounding them (Fig. 3B). Only 59 % of the cells (N = 384, in 3 experiments) were identified as positive.

To optimize our culture medium, we tested several FCS (Gibco; Seromed; Ackermann), one serum substitute (Ultroser G, Gibco) and one serum-free chemically defined medium (Loo & Fugnay, 1987). No attachment of the truncal neural tube was observed with the serum-free medium, even on fibronectin-coated dishes. Attachment, but no or weak migration with some FCS batches and reproducible good cell spreading, was obtained with Ultroser G (Fig. 1). We also found that the uPA/tPA ratio and the production of inhibitor-tPA complex was influenced by the batch of serum used (Table 1). Ultroser G was used as medium supplement for the standard NCC isolation procedure.

The activity of the PAs secreted by NCC is correlated with the number of cells in the culture. The mean activity per cell in five cultures was 0·92 ± 0·27 μPlough for tPA and 1·20 ± 0·46μPlough for uPA. The activity of the inhibitor-tPA complex is 0·53 ± 0·17 μPlough.

The dissected truncal neural tube was subdivided into four equal segments extending from the last 6 somite pairs to the neural plate area adjacent to Hensen’s node (Fig. 4) and cultured for 48 h, after which the neural tube explant was removed. After a further 12 h of culture of the NCC in serum-free medium, the supernatant was examined by electrophoresis and by zymography. Fig. 4 shows the production pattern of PAs along the neural tube segments. The production of total tPA (tPA and inhibitor-tPA complex) was fairly constant along the truncal part of the neural tube and neural plate, whereas uPA activity was highest in the last segment. No morphological difference was observed in the cultured cells. The control unsegmented truncal neural tubes had PA activities similar to the sum of the 4 segments.

The presence of PA in mammalian NCC was suspected since Valinsky & Le Douarin (1985) have demonstrated the production of a plasminogen-dependent protease in quail NCC cultures. This study shows that mouse truncal NCC do secrete PAs and that both urokinasetype and tissue-type enzymes are produced. The identification of the two forms of PAs was confirmed by comparison with references and by immunoprecipitation.

The localization and the identification of quail NCC is made easy either by their morphology or by their ability to undergo melanogenesis (Cohen & Konigsberg, 1975), or by the detection of specific markers on their cell surface, such as HNK-1 or R24 for example (Maxwell et al. 1988). Mouse NCC are less well-defined so far (Simonneau et al. 1987). The presence of melanocytes or even pre melanocytes, after DOPA reaction (Hirobe, 1988), has been observed only sporadically in culture. Therefore, we examined our cultures for the presence of NCC by reintroducing the isolated cells into developing embryos. The results obtained are in accord with those of Jaenisch (1985) and confirm that at least part of the cells we are working with are genuine NCC that have kept their potential to differentiate into melanocytes after 2-3 days in culture. The observation of catecholamine-positive cells also confirms the presence in the cultures of NCC capable of undergoing a neuronal pathway. Although freshly isolated truncal neural tubes were carefully examined for the absence of contaminant non-neural tissue, and although the neural tube expiants were dissected out after 48 h of culture, leaving the stellate cells, we cannot exclude the presence of non-neural crest cells (Loring et al. 1988). It should be kept in mind that what we call NCC include all the cells that show early migratory properties in vitro (Le Douarin, 1982). The high sensitivity of NCC to external factors in vitro led us to set up standardized culture conditions based on Ultroser G, a serum substitute, in MEM.

It has been shown that cultures of early-emigrated quail NCC were composed of a heterogeneous cell population (Le Douarin, 1986). In this context it would be of interest to determine whether all mouse NCC or a subpopulation only produce PAs. Although this alternative could not be tested in primary cultures, the results obtained with secondary cultures indicate that about half of the cells have a detectable PA activity. This is in favour of the heterogeneous population hypothesis and could be verified in a more direct manner by localizing the mRNA for uPA and tPA by in situ hybridization (Debrot & Menoud, in prep.).

NCC need to disrupt several barriers to migrate. First, they have to cross the basement membrane of the neurectoderm, then their migration within the embryo takes place through the ECM. When the first NCC start to migrate, before neural tube closure, the basement membrane of the neurectoderm must be disrupted (Menoud et al. 1989; Sternberg & Kimber, 1986). The very early disappearance of type IV collagen, fibronectin and laminin, three components of basement membranes and ECM, when NCC migration onset is supposed to occur, suggest a high proteolytic activity. This correlates with our observation of an intense secretion of uPA in the most caudal segment. On the other hand, the tPA secretion appears constant along the truncal neural tube and its role in migratory processes might be different and dependent on its regulation by inhibitors.

Though the secretion of tPA and uPA by NCC has been demonstrated, a confirmation of their presence in vivo and an insight into their regulation are required to obtain a better understanding of the significance of these enzymes in NCC migration.

We thank Alice Zosso for her excellent technical assistance and Professor J.-D. Vassalli for his valuable comments on the manuscript. This work was supported in part by a grant from the Sandoz Foundation to S. D.