Mitogenic activity of chicken chorioallantoic fluid is temporally correlated to vascular growth in the chorioallantoic membrane and related to fibroblast growth factors

The chorioallantois of the chicken embryo originates from a small diverticulum protruding from the tail gut into the chorion cavity at the second day of development. This diverticulum, the allantois, consists of an outer vascularized mesodermal and an inner endodermal layer. At day 4 the mesodermal layer fuses with the chorionic membrane lining the shell membrane and the allantois acquires an ectodermal coat and is subsequently called chorioallantois (Hamilton, 1965).

The choriollantois is essential during embryonic development. As it includes the chorioallantoic fluid (CAF) into which waste products are delivered, it can be regarded as an ‘in ovo urinary bladder’. Additionally, serving as the gas exchanging surface, it resembles an ‘in ovo-lung’ (Hamilton, 1965; Romanoff, 1967).

The respiratory function is provided by the extensive capillary network of the chorioallantoic membrane (CAM). Growth and differentiation of its capillaries are closely associated with the development of the chorioallantois. Capillaries continue to grow up to day 11 and then proliferation of endothelial cells ceases, coinciding with an almost completed extension of the chorioallantois (Ausprunk et al. 1974).

Over the past decade, it has become evident that developmental events are controlled by polypeptide growth factors. Fibroblast growth factors (FGFs) constitute a family of strongly conserved heparin-binding mitogens which are supposed to be of major importance in angiogenic processes (for review see Baird and Walicke, 1989). Indeed, basic and acidic FGF could be identified in embryonic tissues of chicken brain and kidney and were shown to be temporally correlated with vascularization of these organs (Risau, 1986; Risau and Ekblom, 1986; Risau et al. 1988a). Both factors possess potent mitogenic activity for endothelial cells as well as a broad spectrum of mesodermal and neuroectodermal cells (for review see Gospodarowicz et al. 1986; Burgess and Maciag, 1989). A crucial developmental role of fibroblast growth factors is further substantiated by their ability to induce mesoderm formation (Slack et al. 1987) and to regulate myoblast differentiation (Clegg et al. 1987). In this context, the presence of bFGF within embryonic sceleton muscle cells is an interesting phenomenon (Joseph-Silverstein et al. 1989).

Since the CAM represents one of the densest vascular networks in the chick embryo, it can be assumed that angiogenic factors are involved in chorioallantoic vascular growth. We hypothesize that these putative factors are at least partly released into the chorioallantoic fluid (CAF). Since furthermore CAF is much more easily available than CAM extracts, we analyzed CAF for presence of angiogenic and mitogenic activities. The biological activities are characterized and compared with the growth-promoting effects of other embryonic fluid compartments such as amnion fluid and embryonic serum. Furthermore, since angiogenesis of the CAM is strongly correlated with certain developmental stages, the temporal distribution of growth activities is examined.

Fertilized White Leghorn eggs were incubated at 37–38 °C in 60% relative humidity. Chorioallantoic fluid (CAF) was collected from embryos at days 6 to 14 of incubation, amnion fluid and serum at day 9. After disinfection, the egg shell and shell membrane were removed over a wide area of the chorioallantois avoiding any injury to the CAM. CAF was soaked up using a syringe with a fine gauge. In the same way, amnion fluid was collected by puncturing the amnion cavity through the chorioallantois. Serum was obtained by slitting a large chorioallantoic vessel using an electrolytically sharpened tungsten needle (Dossel, 1958) without injuring the CAM. Released blood was soaked up from the surface of the CAM. All fluids were cleared by centrifugation and supplemented with antibiotics (Serva) (final concentrations: 1000i.u.ml⁻¹ pénicilline, 0.1 mg ml⁻¹ streptomycine). In most of the CAF samples contamination by serum was inevitable, but did not exceed 1 %. All serum samples were free of CAF.

Chorioallantoic fluid (CAF) of 9 day embryos was concentrated 15-to 20-fold using centricon-10™ microconcentrators (Amicon Corp., Danvers, USA). The concentrate was either mixed with 4 % agar at 40 °C or with 1 % methylcellulose in a ratio 1:1. The agar mixture was cast into a glass tube and, after solidification, the agar rod was cut into discs of equal size. Methylcellulose discs were prepared by pipetting 40 μl of the mixture on sterile glass slides. Solidified discs were implanted onto the chorioallantoic membrane (CAM) of 9 day chick embryos. For control experiments PBS was applied instead of CAF. Three days after reincubation, the experiments were evaluated. In one group of experiments, the vascular system of the embryos was injected with Indian ink (Christ et al. 1973) in order to make small blood vessels visible. The exposed areas were fixed in Serra’s fluid, dehydrated in a graded series of alcohol and clarified in methylbencoate. A second group of experiments was processed for light microscopy. Exposed CAM areas were fixed in 4% glutaraldehyde in 0.12 M cacodylate buffer (pH 7.4) and embedded in Durcupan® after dehydration in a graded series of acetone. Semithin sections were cut transversely through the whole exposed areas and stained in 1% methylene blue. In a third group of experiments, exposed areas were analyzed under a dissection microscope for signs of angiogenesis. Angiogenic reactions were evaluated according to the following criteria: positive response: ‘spoke wheel’ pattern of blood vessels with newly formed small blood vessels in its center; questionable response: ‘spoke wheel’ pattern of blood vessels without newly formed blood vessels in its center; negative response: no reaction, edema and vascular deviations other than ‘spoke-wheel’-like.

The mitogenic activity of CAF, amnion fluid and serum was determined using Swiss 3T3 cells as target cells. Cells were seeded in 96-well plates at a density of 7.5 times 10³ and grown in 0.1ml DMEM supplemented with 0.5% fetal calf serum (FCS) and antibiotics. After 16–24 h 25 μl of fluid samples was added. Following 24 h incubation, cells were labelled with [³H]thymidine (Amersham) at 0.25 μCi/well. After another 24 h, cells were harvested on glass fiber filters. Incorporation of radioactivity was measured by scintillation counting and determined as average of 5 wells.

10 ml CAF was loaded onto a heparin-sepharose column (0.5×5cm) equilibrated in 10mM Tris-HCl pH8.1. After extensive washing, bound proteins were eluted with a 30 min linear gradient of 0–2.7 M NaCl in the same buffer. Fractions of 1 ml were collected, equilibrated in DMEM and 0.5 % FCS and subjected to mitogenicity assays. For Western blot analysis, proteins were eluted at steps of 1.2 and 2.4M NaCl.

Fractions obtained by heparin-sepharose affinity chromatography, CAF and serum were desalted in 50 mM ammonium bicarbonate on Sephadex G-25 and lyophilized. Lyophilized recombinant human bFGF was obtained from Progen, FRG. Samples were redissolved in sample buffer (0.1M Tris-HCl pH6.8, 1% SDS, 10% glycerol, 5% mercaptoethanol) and separated on 12.5–17.5% linear gradient polyacrylamide gels. Proteins were transferred electrophoretically on PVDF membranes (Immobilon, Millipore) in 50 mM Tris-HCl pH 8.3. Blots were blocked for Ih with 2% defatted milk powder in Tris-buffered saline pH 7.4 and 0.5% Tween 20 (TBST) and incubated for 2h in anti-bFGF antiserum (Schweigerer et al. 1987) diluted 1:1000 in TBST. Bound antibody was detected with alkaline phosphatase-conjugated second antibody (Dianova, Hamburg, FRG) using 5-bromo-4-chloro-3-indole phosphate and nitroblue tétrazolium as substrates.

Immunoglobulin fractions from rabbit anti-bFGF antiserum and a non-immune serum were isolated by affinity chromatography on protein A sepharose (Pharmacia). Crude CAF was preincubated with varying concentrations of the immuno globulins for 1 h at 37°C and then subjected to 3T3 mitogenicity assay as described above.

The rich vascularization of the CAM suggests that angiogenic substances are involved in the regulation of chorioallantoic vascular growth. It is supposed that these substances may also be detectable in the chorioallantoic fluid (CAF). Consequently, activities accumulated in the CAF should reflect the temporal changes in growth-promoting stimuli in the CAM and probably correlate with changes of chorioallantoic vascular growth.

As a homologous (chick–chick) in vivo screening test for angiogenic activity, we used the chorioallantoic membrane assay (CAM assay). CAF of 9-day-old embryos was applied to the CAM of 9-day-old embryos and found to induce an angiogenic response of high reproducibility. After three days of reincubation, an angiogenic reaction was observed beneath the disk in 78% of the experiments (n=32). The reaction comprised ‘spoke wheel’ transformation of the chorioallantoic vascular pattern with an extensively vascularized center. As revealed by injection of Indian ink, this vascularized center consisted of a sponge-like network of newly formed small blood vessels, which were supplied with blood and drained by vessels of the ‘spoke wheel’ arrangements (Fig. 1A). In 19% of the experiments, angiogenic reactions were restricted to a ‘spoke wheel’-like transformation of the vascular pattern without any formation of small vessels in the center. This type of response was subsumed under the category ‘questionable response’. A negative response was obtained in only 3%. For comparison, 63% of the control experiments (n=43) gave a negative (Fig. IB), 24% a questionable and 13% a positive response.

Further analysis of angiogenic and negative responses was obtained by light microscopy of semithin sections (Fig. 2). The sponge-like bundle of small blood vessels corresponded to a mesenchymal reaction that enclosed both the augmentation of fibroblasts and the formation of numerous capillaries (Fig. 2A). Negative control experiments, on the other hand, exhibited edema without additional vessel growth (Fig. 2B). Moreover, in CAF, as well as in control experiments, thickening of the chorioallantoic ectoderm and endoderm, infiltration of the interstitium by inflammatory cells and exulcerations beneath the carrier could be observed. Similar effects were previously described by Jakob et al. (1978) and Spanel-Borowski et al. (1988) as inflammatory reactions, which can be induced by different carrier materials. These effects, however, were more pronounced in CAF than in control experiments.

We further analyzed whether the angiogenic activity of CAF observed in vivo coincides with a mitogenic activity in.vitro. CAF of day 9 was tested in 3T3 cell proliferation assays and compared to activities of serum and amnion fluid. As compiled in Fig. 3, both CAF and amnion fluid promote cell proliferation, with CAF being three times more potent than amnion fluid.The activity of both fluids was slightly enhanced at dilutions between 5- and 10-fold and decreased at higher dilutions. In contrast, native serum of day 9 embryos had no proliferative effects and even inhibited [³H]thy-midine incorporation.

Since capillary growth of the CAM progressively occurs up to day 10 and thereafter decreases, we questioned whether mitogenic activities can be correlated with the kinetics of vascular growth. Therefore, CAF from day 6 to 14 was collected and examined in mitogenesis assays. As pointed out in Fig. 4, high proliferative capacity is detected in CAF from day 6 to 9. A maximum at day 7 is followed by a slight decrease to day 9. From day 9 to 10 biological activity sharply decreases reaching a plateau or slight decline from day 10 to 14. The sharp regression of mitogenic activity occurs one day before vascular growth of the CAM is reduced.

For further characterization of the mitogenic activity, CAF was separated by heparin-sepharose chromatography, a method suitable for the selective purification of fibroblast growth factors. Fractions collected were tested for their ability to stimulate 3T3 cell profiferation.

Indeed, as shown in Fig. 5, most of the bioactive material was adsorbed by the column. Maximum activity was recovered in two distinct peaks eluted at 0.8–1.6 and 1.6–2.6M sodium chloride, respectively.

This elution profile strongly indicated the presence of fibroblast growth factors. To confirm this assumption fractions eluted from heparin–sepharose at 1.2 and 2.4 M NaCl were separated by SDS–PAGE and analyzed in Western blots with a monospecific antiserum against bFGF. (An immunological detection of aFGF is in preparation). Western blot analysis with the anti-bFGF antiserum revealed an immunoreactive protein of about 17×10³M_r in the 2.4 M fraction (Fig. 6). This protein was more clearly detectable in native CAF (Fig. 7). It is somewhat larger than recombinant human bFGF (Progen, FRG) migrating at about lóxllPAfr. The bFGF-like protein was confined to CAF and not detectable in embryonic serum, which - as shown above - lacked any mitogenic potential. In addition, both CAF and scrum contained a protein band of about 60×10³M_r. This immunoreactive protein is probably nonspecifically recognized by the antiserum (Fig. 7).

To determine whether mitogenic activity of the CAF is caused by the bFGF-like immunoreactive protein, an immunoinhibition assay was conducted. Native CAF was incubated with the immunoglobulin fraction from the rabbit anti-bFGF antiserum and a control immunoglobulin at varying concentrations and was then tested for mitogenic activity on 3T3 cells. Incubation with the anti-bFGF immunoglobulin remarkably blocked the mitogenic activity of the CAF in a dose-dependent manner. At highest immunoglobulin concentrations, activity was reduced for more than 50%. The control immunoglobulin had no effect on mitogenic activity (Fig-8).

In the present study, chorioallantoic fluid (CAF) of chick embryos was examined for the presence of angiogenic and mitogenic activities. Four main conclusions can be drawn from our results. First, mitogenic activities in CAF are probably due to fibroblast growth factors. This is suggested from chromatographic behavior of CAF after heparin-sepharose chromatography. The two heparin-binding activities obtained in CAF clearly correspond to elution profiles reported for aFGF and bFGF, respectively (Lobb et al. 1986; Risau et al. 1988a). The identity of bFGF in CAF was further substantiated by Western blot analysis with a monospecific antiserum. (An immunological detection of acidic FGF is currently under investigation.) Immunoinhibition assays with the anti-bFGF antibody clearly demonstrated that bFGF is a main reason for mitogenic activity in the CAF. Second, comparison of different embryonic fluids by biological and immunological criteria revealed that bFGF is accumulated in the CAF but not shifted from serum into the CAF. Third, temporal expression of CAF mitogenic activities strikingly corresponds to vascular growth patterns observed in the CAM (Ausprunk et al. 1974). Fourth, CAF is strongly angiogenic in vivo. This result from chorioallantoic membrane assay is in line with previous studies that demonstrate that FGFs elicit strong angiogenic responses when applied to various in vivo systems (Gospodarowicz et al. 1979; Risau, 1986; Thompson et al. 1989).

Fibroblast growth factors have been isolated from a variety of adult tissues. Whereas acidic FGF is generally restricted to neural tissue, basic FGF seems to be more widespread (for review see Gospodarowicz et al. 1986; Burgess and Maciag, 1989). FGFs possess a molecular range of about 16–18×10³M_r. Recent studies reveal evidence for the presence of additional high molecular mass forms in the range of 22–25×10³M_r. Differences in molecular weights are presumably due to different translational initiation or truncation of N-extended forms (review by Rifkin and Moscatelli, 1989). It has been reported that commonly applied acidic conditions may account for N-terminal cleavage of FGF (Klags-brun et al. 1987). As these conditions, however, were avoided in the present study, it is conceivable that the 17–10³M_r immunoreactive bFGF-like protein represents a native and untruncated form. This protein was not detectable in embryonic serum.

Both aFGF and bFGF bind tightly to heparin and, probably more important, to heparan sulfate. Indeed, recent biochemical and immunohistological data have demonstrated that FGFs are closely associated with the extracellular matrix of their cellular sources (Vlodavsky et al. 1987; Folkman et al. 1988; Gordon et al. 1989; Presta et al. 1989). Hence, it is rather surprising that in the CAF the FGFs are apparently present in free solution.

However, the presence of FGFs, probably related to aFGF and bFGF, in the CAF, is analogous to the presence of both factors in the vitreous of the chick embryo, which is also an acellular fluid compartment. In this compartment, FGFs may have their function in induction of the proliferation of lens epithelial cells (Mascarelli et al. 1987).

Since genes of aFGF and bFGF do not encode a typical signal sequence, the growth factors are probably not released by conventional secretory pathways (for review see Burgess and Maciag, 1989). In tissue repair and inflammatory processes in which FGFs are presumably implicated, it has been suggested that growth factors are released by cell lysis or hydrolytic enzymes from extracellular structures (Gajdusek and Carbon, 1989; Schulze-Osthoff et al. 1990). In the chick embryo, however, FGF release cannot be explained by cell death. Moreover, it is possible that growth factors are released into the chorioallantoic fluid by specialized transport mechanisms.

The expression of FGFs in CAF closely coincides with the progression of vascular growth in CAM indicating the role of these growth factors in blood vessel formation. In this respect, it is interesting that the observed decrease in CAF mitogenic activity per volume strikingly corresponds to the regression of vascular growth in the CAM (Ausprunk et al. 1974). In Western blot analyses, we found that the decrease in mitogenicity is probably not due to differential synthesis of inhibitors, but to a decrease in bFGF immunoreactive protein during development (data not shown). Changes in mitogenic activity and immunoreactivity, however, do not correlate with the changes in total CAF volume, which attains a maximum at day 13 of development (Romanoff, 1967). The time interval between decrease of mitogenicity and regression of capillary growth may reflect the period over which released FGFs remain active.

The correlation between changes in the production of FGF and vascular growth activity, which we describe here, is to our knowledge the first in vivo correlation between decrease of FGF expression and regression of capjllary growth.

In contrast, previous studies in embryonic development mainly describe a correlation between increasing levels of FGF and induction of vascular growth, including studies on central nervous system (Risau, 1986; Risau et al. 1988a), embryonic kidney (Risau and Ekblom, 1986) and embryoid bodies (Risau et al. 19886). In these studies, both angiogenesis, i.e. formation of new blood vessels by immigration of endothelial cells in primarily avascular organs, and vasculogenesis, which means in situ differentiation of a vascular plexus from presumptive endothelial cells, were correlated to the expression of aFGF and bFGF related mitogens.

The contribution of FGFs in embryonic angiogenesis was further supported by immunohistological means. Hanneken et al. (1989) showed that newly formed capillaries of the bovine fetal retina stained positively with bFGF-specific antibodies. In our immunohistochemical investigations, we were not able to identify a cellular source of the bFGF-like immunoreactive protein of the CAF. CAM itself as well as the embryonic kidney which contributes to the chorioallantoic fluid (Boyden, 1924) and recently was shown to produce a FGF-related factor (Risau and Ekblom, 1986), are potential sources of the aFGF- and bFGF-related activities.

Whether FGFs are the only agents in the native CAF inducing angiogenesis can not be decided from our experiments. As angiogenesis is thought to be a process of multifactorial genesis (Folkman and Klagsbrun, 1987), other diffusible angiogenic factors may be released into the CAF additionally to FGFs.

Recently, further FGF-homologous oncogenes have been identified, which, as far as presently established, display similar biological activities as aFGF and bFGF, but, most remarkably, contain functional secretory signal sequences (for review see Burgess and Maciag, 1989). Despite a still limited number of investigations, the expression of these additional members of the FGF family seems to be confined to the early embryonic development whereas no transcripts can be detected in adult tissues. K-FGF was found to be expressed in embryonic stem cells (Heath et al. 1989). The expression of the int-2 proto-oncogene has recently been studied in the early mouse embryo and its pattern of localization suggests possible roles in cell migration during gastrulation and neurulation (Wilkinson et al. 1988) .

As far as these members of the FGF family are concerned no data exist about their expression in chicken embryos. Studies of their distribution in mammalian species, however, suggest these factors are suitable candidates for inductive or cell specification processes. Further comparative studies on the temporal and spatial expression of several FGF members should therefore be promising in elucidication of embryonic blood vessel growth.

The authors wish to thank Dr L. Schweigerer, Heidelberg, for kindly providing the anti-bFGF antibody.