Tumor Angiogenic Factor PURIFICATION FROM THE WALKER 256 RAT TUMOR*

We report here the purification of an angiogenic substance from the Walker 256 rat ascites tumor which is mitogenic for fetal bovine aortic endothelial cells in culture and which also stimulates new blood vessel growth in vivo. Purification was monitored in endothe- lial cell cultures by cell counting or by [‘I-IJthymidine incorporation into DNA. Lyophilized crude tumor cell homogenate was extracted with absolute ethanol, and the extract was further purified by silica gel column chromatography. The most purified material eluted with ethyl acetate:methanol (26:l) and behaved as a single component when analyzed either by thin layer chromatography with both silica gel and polyethylene- imine-cellulose systems or by reversed phase (Cia) high pressure liquid chromatography. Bio-Gel P-30 chro- matography indicates a M, 400-800 for the active material. The mitogenic activity of the purified material was observed with the vascular endothelial cells cultured in serum-less as well as serum-containing media. The angiogenic activity of this material was revealed in both the chicken chorioallantoic membrane and rat corneal micropocket bioassays. Our results demon- strate that a factor with mitogenic and angiogenic activities can be purified from the Walker rat tumor by a novel, facile, and high yield process. The sig-dicance

The sig-dicance of the interaction between host vacdature and solid malignancies to the survival and growth of tumors has been recognized for over a century. Both Virchow (1) and Thiersch (2) described the altered vascular network about tumors. Ribbert (3), Goldmann (4), and Russell (5), in separate studies, noted the irregular and tortuous vessels about the tumor and in the adjacent tissue, and concluded that such findings were useful as an indicator of the presence of a malignancy. In fact, Goldmann observed further that tumor cells determined vascular proliferation and that the newly developed vessels promoted tumor growth (4).
Studies on the dynamics of this interaction between tumor and host were advanced by the development of the transparent chamber method in a variety of experimental animal models (6,7). From these studies arose the opinion, expressed early by Ide et al. (8), "that there is associated with the viable growing tumor cells some blood vessel growth stimulating factor." A clear-cut demonstration of the existence of such a hmoral, angiogenic factor was first provided by Greenblatt and Shubik (9), in their transfilter d i i i o n studies. Elucidation of the exact nature of this angiogenic factor has not yet been achieved. In an early report by Folkman et al. (ID), the factor from the Walker 256 rat tumor was isolated and characterized as a relatively large substance containing ribonucleic acid, protein, and carbohydrate. Kumar, Weiss, and their associates proceeded from this stage in the purification and eventually isolated a more highly purified angiogenic agent that had an apparently low molecular weight (-200) and was not composed of protein, peptide, nucleic acid, or prostagIandin (11). The isolation from the Walker rat tumor of a low molecular weight angiogenic factor has subsequently been claimed by McAuslan and Hoffman (12). None of these studies has produced any evidence relating to the purity of their isolated angiogenic factor(& or an unequivocal structure determination for this factor.
In the following we present the novel procedures by which we have purified to homogeneity an angiogenic factor derived from the Walker 256 rat tumor. Since these studies were facilitated by the use of a cell culture assay based on growth effects, we also describe in detail this methodology.

MATERIALS AND METHODS
Cell Culture and Assay Conditions-The procedures for obtaining and maintaining cultures of fetal bovine aortic endothelial cells have been previously described in detail (13,14). M i n i u m essential medium (Grand Island Biological Co., Grand Island, NY) containing Earle's sal ts, peniciUii (200 units/ml), streptomycin (200 pg/ml), glutamine (2 mM), and fetal bovine serum (IO%, Sterile Systems, Inc., Logan, UT) was employed as the growth medium. Subculturing was routinely performed every 2 days using split ratios of 1:3 to 1:5. Generally, cells from passages 10 to 18 were used for testing purposes. Fetal bovine aortic smooth muscle cells and adult bovine corneal fibroblasts were isolated by standard methods (14), maintained in culture, and placed on test using conditions identical with those used for the endothelial cells.
Cell growth studies were carried out in 24-well plastic dishes (Falcon). Cells for testing were treated as in a typical subculturing, except that cell counts were determined with a Coulter counter. Using these conditions, cell viability of 295% had been obtained by hemacytometer measurements of cells suspended in a trypan blue solution. Cells resuspended in the growth medium were plated in the multiwell dishes at the desired concentration; 5 or 16 h later the attached cells (approximately 60-75% of the original plant) were rinsed with Caand Mg-free phosphate-buffered saline solution. Either of two types of test media (at a volume of 1 ml/well) was then employed, serumless Medium 199 or serum-containing Medium 199. Serum additions to media were either 1-2% fetal bovine serum or 2-2.5% dialyzed FBS.' See figure legends for specific serum conditions. Test materials were generally added as Hz0 solutions whose volumes never exceeded ' The abbreviations used are: FBS, fetal bovine serum; HPLC, high pressure liquid chromatography; TLC, thin layer chromatography.

9606
Purification of a Tumor Angiogenic Factor 0.10 ml. Replicate wells on the same dish were employed for any given test condition. The actual number of cells present at the start of the test was measured by detaching the cells from several wells with 0.1% trypsin (Type I11 from bovine pancreas, Sigma) and 0.05% EDTA, and carrying out cell counts. At 5 h attachment efficiencies of 60-75% were obtained; slightly higher cell counts were obtained after an overnight (16 h) incubation in growth medium.
Test cultures were incubated at 37 "C in a humidified 5% COz atmosphere for the specified period of time. At the end of this period, the cultures were processed for quantitation by cell counting or incorporating tritiated deoxythymidine into cellular DNA. Cell counting was performed with a Coulter counter on cells detached with a trypsin-EDTA solution. Incorporation of [3H]deoxythymidine was achieved by the following pulse-labeling technique; 1 h prior to the end of the experiment the test media were removed and replaced with 1.0 d/well of serum-free minimum essential medium containing 0.625 pCi/ml of [3H]dThd ([methyZ-'H]thymidine obtained from Amersham (Arlington Heights, IL) as a sterile, aqueous solution of 20-25 Ci/ m o l , diluted to 6.7 C i / m o l and used within 2 months). After a 1-h incubation at 37 "C the incorporation was terminated by removing the culture fluid and rinsing each well once with cold (4 "C) Ca-and Mg-free phosphate-buffered saline solution. This was followed with a 5-min rinse at room temperature with a solution of 50% Ca-and Mg-free phosphate-buffered saline solution and 50% acetic acidethanol (1 part glacial acetic acid and 3 parts 80% ethanol). The cells were fixed to the plastic surface and extracted for a period of 30 min to 3 h with the acetic acidethanol solution. After fixation the cells were re-extracted with 0.3 N perchloric acid for 10 min and then thoroughly rinsed with H20. Solubilization of cellular materials was accomplished by incubating the contents of a well with 0.2 ml of 0.25 N sodium hydroxide solution for 20 min at room temperature. The contents of each well were transferred to 4-ml glass counting vials using an additional 0.2 ml of HzO for rinsing the well. Each vial then received 3.2 ml of counting fluid (Budget-Solv, Research Products International, Elk Grove Village, IL), the contents were mixed, and the vials were counted in a Beckman LS-100 liquid scintillation counter (with a tritium-counting efficiency of approximately 31%). All data are minimally the average of duplicate samples. Separate autoradiographic studies performed on cells labeled and fixed in an identical manner revealed that the tritium label was localized to the nuclear regions of the cells.
Tumor-derived Materials and Chromatographic Techniques-Walker 256 ascitic tumor cells, carried in male Sprague-Dawley rats, were collected, centrifuged at 800-1000 X g for 10 min to yield a tumor cell pellet, washed two or three times in 0.87% (w/v) ammonium chloride to lyse the red blood cells, and recentrifuged. The nearly white cell pellet was resuspended at 80 X 10' cells/ml in 50 mM Tris-HCl (pH 7.3). The suspension was sonicated at 100 watts for 4 min at 4 O C ; the sonicates were centrifuged (25,000 X g for 20 min at 4 "C). The active supernatant, termed crude tumor homogenate, could be stored frozen at -20 "C for months with little loss in activity.
Crude tumor homogenate (about 1 g of Lowry protein or 4500 A2W -) was lyophilized and extracted with absolute ethanol (producing an ethanol extract). Best recoveries of active material (with a total of about 700-800 AZm -) were obtained by sonicating (for a total of three times) a finely divided suspension of freeze-dried tumor cell homogenates in 100 ml of ethanol for a total of 2 min at 4 "C and 100 watts. Extracts were centrifuged (21,000 X g for 15 min); the supernatants were pooled and taken to dryness using a rotary evaporator. The residue was redissolved in either methanol for silica gel chromatography or in aqueous solutions for Bio-Gel chromatography or for assaying endothelial cell growth stimulatory activity. For comparative studies tissue homogenates and the corresponding ethanol extracts were prepared from rat liver, kidney, and skeletal muscle, using the same procedures as have been described for the Walker rat tumor cells.
Bio-Gel P-30 (Bio-Rad) column (2.7 X 25 cm) chromatography was performed. Molecular weight markers were chymotrypsinogen (M, were applied to the column and eluted with 50 mM Tris-HC1 (pH 7.0). Chromatography was also performed at pH 3.9. Acid treatment of crude tumor homogenate was carried out by adjusting the pH to 3.9, using acetic acid, and then centrifuging the suspension at 45,000 X g for 20 min at 4 "C. The inactive pellets were discarded; 4 ml (18 AZW , of the supernatant were applied to the column and eluted with 50 mM formic acid at pH 3.9. In all cases the flow rate was maintained at 0.5 ml/min; 1.0-ml fractions were col-lected. The absorbance of the eluate was monitored at 280 and 254 nm using an ISCO model UA 5 monitor/recorder (Instrumentation Specialties Co., Lincoln, NE); the absorbance of individual fractions was determined with a Gilford spectrophotometer. Peak fractions were sterilized by filtration (0.2 pm Millipore Nters) and tested for activity directly. In addition, pooled fractions from the pH 3.9 chromatography were taken to dryness by lyophilization, redissolved in water, sterilized by filtration, and placed on test.
Silica gel (column chromatography grade, E. Merck AG, Darmstadt, West Germany) was exhaustively washed with HPLC-grade methanol (Fisher) and dried by vacuum desiccation at 40 "C before use. Silica gel chromatography of the ethanol extract was carried out using a column (2.5 X 36 cm) initially equilibrated with chloroform.
The sample was applied in 15 ml of methanol at about 23 AZM) ,,,,,/ml. Generally, the sample volume was less than 10% of the column volume and approximately 2 Am , , , , , were applied per ml of column volume.
Using a flow rate of 5 ml/min, 7.5-ml fractions were collected. Elution was effected using five solvent steps: chloroform, ethyl acetate, ethyl acetate:methanol (3:1), ethyl acetate:methanol (l:l), and methanol, in the order stated. All solvents used were HPLC grade.
Rechromatography on silica gel was performed with active material that eluted with ethyl acetate:methanol(3:1). This fraction (7.5Am in 1 ml of methanol) was placed on a silica gel column (1.2 X 9 cm) packed in chloroform. Stepwise elution at a flow rate of 1 ml/min was carried out with chloroform, ethyl acetate, and mixtures of ethyl acetate:methanol in the ratios of 501, 251, 101, 5:1, and 21, in the order stated. The absorbance at 260 nm was measured with a Gilford spectrophotometer for each of the 1.0-ml fractions collected.
Thin layer chromatography, using both silica gel and polyethyleneimine-cellulose adsorbents, was carried out on Brinkmann plates (with fluorescent indicator) in closed glass chambers. Samples of 10 pl were spotted in methanol using glass capillary tubes. Silica gel plates were developed with methanol; ion exchange plates, with 1 M NaCl in 0.1 M Tris-HC1 (pH 7.0). The components were routinely visualized using three methods: 1) fluorescence quenching; 2) iodine adsorption; and 3) reaction with ninhydrin (15). Silica gel plates on occasion were treated with Zinzadze reagent, bromthymol blue, or charring after spraying with sulfuric acid (15); however, no additional components were detected with these reagents other than those observed with the other three methods.
High pressure liquid chromatography was performed using a Waters system with an analytical Whatman OD& reversed-phase column (column volume, 2 d). The components were eluted isocratically with 5% aqueous methanol at a flow rate of 1 ml/min. Eluants were routinely detected spectrophotometrically by means of absorbance at 260 nm and occasionally by the alternative means of refractive index differences. Results from these two detection methods were always in excellent agreement.
Vascularization Assays-Neovascularization was examined on the chicken chorioallantoic membrane using the previously described techniques (16). Corneal neovascularization in rats (17) was studied using essentially the procedures employed with rabbit (18) and mouse corneas (19). Briefly, corneal micropockets were prepared in the eyes of male Sprague-Dawley rats; an Elvax pellet containing the test material was inserted into the pocket, which was sealed by repositioning the temporarily displaced corneal epithelium. Every 4 days eyes with these implants were examined using a Zeiss slit lamp stereomicroscope (Carl Zeiss, Inc., New York, NY). Maximum vessel length and the pellet-limbus distance were measured with an ocular micrometer. The number of blood vessels and the presence of any corneal edema were also noted. At the end of the test (day 16), histological studies on enucleated and formalin-fixed eyes were performed using paraffin sections stained with hematoxylin and eosin (17).

RESULTS
In order to simplify and accelerate certain steps in our assay technique, without sacrificing the reliability and flexibility of our previously described procedures (13), we determined conditions for an endothelial cell growth assay based upon the incorporation of radiolabeled thymidine into cellular DNA.
The primary criterion for judging the acceptability of any experimental condition was the close agreement between cell counts and r3H]dThd incorporation for parallel test cultures. Two major considerations emerged from these studies: 1) the need to use subconfluent cell cultures, and 2) the need to eliminate artifacts arising from the trypsinization of cells during subculturing. First, the narrow range of cell density (5-50 X lo3 cells/cm2) over which C3H]dThd incorporation into pulse-labeled cells was linear did not prove to be a serious limitation. Even though slightly more than 2 cell doublings are possible during the test period (2-3 days), we normally observe between 1-2 doublings in cell numbers with control cultures grown in Medium 199 containing 2% (or less) FBS. Second, cells for testing must be allowed to remain on the plating medium for a minimum of 5 h before changing to the test medium. The preferred procedure for tests that required 2-3 days in serum-less medium was to use an overnight preincubation in the original plating medium. In order to allow for some growth during the 16-to 20-h period before the start of a test, cells can be plated at a lower cell density (5-10 X lo3 cells/cm2). If this protocol was followed, fewer cells from the control wells detached during the test period.
The dose-dependence plot of Fig. 1 illustrates some of these points. In this experiment, cell growth effects were examined as a function of the medium concentration of impure tumorderived material (obtained by an ethanol extraction of the lyophilized crude tumor homogenate). Stimulation of endothelial cell growth by this tumor-derived material is evident, regardless of the means for monitoring growth effects, cell counts, or [3H]dThd incorporation. The results from these two methods are in excellent agreement. Furthermore, this dose-dependence plot permits a determination of the amount of endothelial cell growth stimulatory activity present in a sample. A unit of activity is defied as the minimal amount of material (on the basis of volume of addition or any other parameter) that is needed to produce a maximal stimulatory effect on endothelial cell growth. From Fig. 1, then, 1 unit corresponds to material with an absorbance of 0.150 at 260 n m (or 150 milliunits at A2mnm). Although there is some degree of variability inherent in measuring activity, it is still possible to employ this definition for gaining an approximation of the recoveries of active material for each step in the purification process.
Deserving of mention at this point are two aspects of the specificity of this in vitro assay. When the ethanol extract of the crude tumor homogenate (as was used in Fig. 1) was tested on two nonendothelial cell types (adult bovine corneal fibroblasts and fetal bovine smooth muscle cells), no growth stimulation was observed under conditions that produced a positive response from the fetal bovine aortic endothelial cells. In addition, if the growth response of these endothelial cells was examined with crude tissue homogenates, or their ethanol extracts, prepared from normal rat liver, kidney, or skeletal muscle, no stimulation of cell growth could be measured, i.e. cell counts (or C3H]dThd incorporation) were within 20% of control values. These observations, in accord with our previous results obtained with crude tumor homogenates and the vascular endothelial cells (13), demonstrate the selective nature of the in vitro assay.
With the development of this convenient and credible assay we began studies on the purification of the endothelial cell growth factor from the Walker 256 tumor. We observed that dialysis of dilute solutions of crude tumor cell homogenates produced a small but real loss in active materials. Activity could be recovered from the lyophilized dialysates, suggesting that the active materials were small molecules. We then studied the elution of activity from a Bio-Gel P-30 column under neutral and acidic conditions. In crude tumor cell homogenates at pH 7.0, most of the activity was found in the flow through, or void volume, fractions (Fig. 2 A ) . Since the crude homogenate could be treated to pH 3.9 with minimal loss of activity, we applied this acid-treated homogenate to the column, which was then eluted at pH 3.9. Under these conditions most of the active materials eluted from the column with an apparent M , 400-800 (Fig. 2B). Additionally, when crude tumor homogenate was adjusted to pH 3.9, then brought back to neutrality and chromatographed at pH 7.0, the active materials also eluted from the same column in the molecular weight range of 400-800.
Purifkation of the active component was next attempted using extraction of crude tumor homogenates with an organic solvent, instead of acid treatment. When lyophilized crude tumor homogenates were sonicated with ethanol, greater than 95% of the total activity was recovered in the ethanol supernatants. This extract was about five times more active than the crude homogenate in our assay system (based on A Z~) . When this ethanol extract was dissolved in a neutral buffer and applied to the Bio-Gel P-30 column, the molecular weight of the active material was also found to be in the range of 400-800 (as in Fig. 2B).

Purification of a Tumor Angiogenic Factor
The mixture from the ethanol extraction of the tumor cell homogenates could be separated further using silica gel chromatography. By applying the sample in ethanol to a silica gel column packed in chloroform and by carrying out the elution scheme indicated in Fig. 3, seven separate fractions could be resolved. Only those fractions eluted with ethyl acetate:methanol (31) displayed activity in the endothelial cell growth assay.
In order to ascertain the composition of the recovered active materials, three analytical systems, two based on the use of thin layer chromatography, and the third on the use of high pressure liquid chromatography, were developed for conveniently but accurately determining the purity of the column fractions. The TLC methods use different adsorbents, silica gel (Fig. 4) and polyet~yleneimine-cellu~ose (Fig. 5), thereby effecting separations on the basis of different molecular properties. In Fig. 4 can be seen results of silica gel TLC of the ethanol extract of the tumor cell homogenates (Lane A ) , as well as the seven fractions from silica gel column chromatography (Lanes B-13). The means of detection routinely employed and shown in this representation include fluorescence quenching, and iodine-and n~n h y~i n -s t a~n g .
Because no one detection method revealed the presence of all of the components in a given sample, only the combination of the routine methods (on a single TLC plate) was considered to provide a fair picture of the state of purity for a sample. Worth noting in the context of the purification procedure, fraction C1 from the silica gel column appeared to be comprised of 3-4 major components by silica gel chromatography (Fig. 4, Lane   D). Corroboration of this finding was obtained by polyethyleneimine-cellulose TLC (Fig. 5, Lane B ) and by HPLC with a reversed-phase CIS analytic column (Fig. 6A). A similar finding resulted when HPLC detection was accomplished by means of refractive index differences, instead of by ultraviolet absorption.
On the basis of these observations, silica gel chromatography was repeated using a shallower solvent gradient. A methanolic solution of fraction CI from the fist silica gel chromatography was placed on a silica gel column (as before), but elution was carried out in stepwise fashion with ethyl acetate to methanol ratios of 50:l down to 2:l. Six fractions containing UV-absorbing material could be separated by this procedure, but only material eluting with ethyl acetate:methanol (25:l) contained endothelial cell growth stimulatory activity. When this material was examined by silica gel TLC (Fig. 4, Lane a, polyethyleneimine-ce~ulose (Fig. 5, Lane C), and HPLC (Fig.  6B), it was found to be essentially homogeneous. Alteration of detection methods and elution conditions did not lead to any different conclusion. B, active fraction C1 from silica gel chromatography (Fig. 3); and Lane C, active material eluting from second silica gel chromatography with ethyl acetate:methanol, 251.

9609
Thus, a homogeneous preparation of a tumor cell-derived growth factor for endothelial cells can be obtained in reasonable yields by this facile, 3-step process. A summary of the process is provided in Table I. Due to our arbitrary definition of a unit of activity, the yields at the various stages of the purification process must be considered only as estimations.
However, that the pwified material is the major (if not only) endothelial cell growth stimulator present in the Walker tumor cell homogenates seems a reasonable conclusion.
We next examine two features of this purified material. 1) its ability to stimulate endothelial cell growth in a serum-less medium, and 2 ) its ability to stimulate vessel growth in two assays for neovascularization, the corneal implant and chicken chorioallantoic membrane assays. The effects of purified tumor factor on aortic endothelial cell growth were studied in Medium 199 containiig no serum, 2.5% dialyzed fetal bovine serum, or 1% FBS (Table 11). The ability of each of these media to support endothelial cell growth, in the absence of added tumor factor, increases in the order given. The addition of these media of maximal stimulatory amounts of purified tumor factor (20 milliunits at A2m) resulted in stimulatory ratios ( Table 11) that decreased in the order serum-less media > media with 2.5% dialyzed FBS > media with 1% FBS. The mitogenic action of the tumor factor is best demonstrated in the serum-less medium, where the contribution to cell growth by serum factors has essentially been eliminated. The inference from these results is that the tumor factor is capable of stimulating cell growth independently of other serum factors.
Also worth noting is that the same results are obtained by cell counts (shown in the table) and by incorporation of r3H) dThd. Furthermore, at concentrations 10-fold higher than  Table 11, the purified tumor factor continued to produce a maximal growth stimulatory effect. These results indicate that an inhibitory (or cytotoxic) agent, present in the crude cell homogenates (13), had been removed during the course of the purification. Thus, in summary, these observations reveal that the tumor factor is capable of initiating cellular DNA synthesis and cell replication for cultured fetal bovine aortic endothelial cells, and that it can act directly on the cells without the intervention of other serum factors.
The angiogenic activity of this tumor mitogen for cultured endothelial cells was examined in the two most commonly used neovascularization assays, the chicken chorioallantoic membrane and the corneal implant assays. In both instances, preparations of the tumor factor were encapsulated in a slow release polymer (Elvax), which was then implanted in the test system (20). Typical results on the chicken chorioallantoic membrane are shown in Fig. 7. In this micrograph the vascularizing responses elicited by a control implant of tumorderived materials inactive in the in vitro assay and an implant containing active tumor-derived materials are being directly compared at day 7 after implantation. A sizable effect, in terms of the number of new vessel loops directed toward the Elvax pellet, is obtained only when material active in stimulating endothelial cell growth is included in the pellet (Fig.  7B). Essentially no effect is seen with the inactive control fractions (Fig. 7A); this vascular pattern is identical with that obtained when only a blank pellet is used.
Implantation of these materials into the rat cornea produced the same result as that with the chicken chorioallantoic membrane assay (Fig. 8 and 9). The micrograph (Fig. 8) shows considerable outgrowth of limba1 vessels only when the corneal implant contained the tumor factor. This response was quantitated for implants containing approximately the same number of milliunits at A260 of inactive tumor material (fraction B from the initial silica gel chromatography), impure but active material (ethanol extracts of the crude tumor homogenate), and the purified tumor factor (Fig. 9). Neovascularization produced by approximately the same number of AZM) ,,,,, was greater with the purified preparation than with the impure material, material inactive in the in vitro assay was also inactive in this in vivo assay. The neovascular response observed in this rat corneal assay was not associated with inflammation, as judged by the absence of corneal edema from stereomicroscopic examination. Furthermore, the inflammatory response, apparent from analyzing the histological sections and characterized by the presence of an occasional lymphocyte or macrophage and rare polymorphonuclear leu- kocyte, was comparable in the three experimental groups (in Fig. 9), and could not be correlated with the amount of vascularization. The results from both the chicken chorioallantoic membrane and corneal assays unambiguously demonstrate that the endothelial cell growth stimulatory factor is also an angiogenic factor.

DISCUSSION
The purification to homogeneity of a low molecular weight angiogenesis factor from the Walker 256 rat carcinoma has been achieved with good recoveries of activity and with only a few facile steps. In fact, preliminary results in this laboratory indicate that the reported 2-step silica gel chromatography may be simplified to just one step by employing a shallower gradient of ethyl acetate:methanol eluants. The purity of this tumor-derived material has been established using two independent TLC and HPLC systems. These convenient analytic techniques, along with the growth assays using endothelial cell cultures, permit a relatively rapid analysis of chromatographic separations of angiogenic agents from the Walker tumor, as well as from other sources. Furthermore, these techniques offer great promise as diagnostic tools in the detection of neovascular disorders, of which tumor vascularization is but one example. For example, the detection of angiogenic activity in various sera, which we have recently accomplished using these techniques, may prove useful in correlating altered serum levels of angiogenic materials with certain neovascular disease states.
The effect of this tumor factor on endothelial cell growth in vitro compared to other growth factors in other cell systems is deserving of further discussion. The cultures of fetal bovine aortic endothelial cells display a direct, selective, and significant growth response to the tumor factor. This observation is in contradistinction to that made with bovine brain capillary endothelial cells (21). In order to observe any growth effects with tumor-derived materials, the bovine cells appear to have stringent requirements for a collagen substratum and added platelets. We have found that cells from fetal bovine aortic endothelium on collagen-coated surfaces respond to the tumor factor as well as do cells on the usual tissue culture plastic surface. Addition of platelets produces in our assays slight growth inhibitory effects. We tend to attribute these effects to the platelet-derived growth factor (22), which when assayed as the purified substance in our system is an inhibitor of fetal bovine aortic endothelial cell growth. Since our assay is straightforward and correlates well with the angiogenic assays, it s e e m ideally suited for carrying out studies on the detection of certain angiongenic agents (in particular, those agents with endothelial cell mitogenic activity), and perhaps on their mechanism(s) of action.
However, a word of caution must be sounded regarding the general use of these techniques in angiogenesis studies. The process of new vessel growth is comprised of at least two separate processes, and possibly a third. For capillary growth to take place, endothelial cell migration and replication are required. In addition, matrix modification, as in basal membrane degradation, may also be needed. Thus, an angiogenic factor may have for capillary endothelial cells a mitogenic activity or a chemotactic activity or both. Azizkhan et al. (23) have recently demonstrated that heparin affects only the movement of cultured bovine capillary endothelial cells and not their rate of growth. The angiogenic activity of heparin, however, was not determined. A low molecular weight angiogenic factor from the Walker tumor has been claimed to be a nonmitogenic, chemotactic factor; however, no supporting experimental evidence was provided (24). The tumor factor purified by us has been examined in our in vitro assay system, which measures only cell growth using endothelial cells from a major vessel, albeit from a fetal source. These same cells have recently been employed in a cell migration assay that has demonstrated the coincidence of mitogenic and chemotactic activities in an angiogenic preparation from bovine retinas (25). Therefore, the relationship between mitogenic and chemotactic activities for the tumor factor should be clarified shortly.
Finally, with the exception of the approximate sue of the pure material (<800 daltons) and the ultraviolet absorption at 260 nm, which was associated with the active material through all the steps of its purification, no other physical characteristics of this substance have been reported here. The methods used for isolating and purifying this material do not permit any unequivocal statements regarding molecular structure. In light of the apparent s m d size and purity of our angiogenic material, it should be amenable to the usual spectroscopic methods of structure determination. The results of these studies will be considered in a future communication.