Eye Res. (1979) 28, 501-514
Eq.
The Angiogenic Activity of the Fibroblast and Epidermal Growth Factor D. GOSPODAROWICZ*~, H. BIALECKI*
AND T. K. THAKRAL,!:
*Cancer Research Institute, Departm,ent of tMedicine wad @Twrgery, lrGvers7Ity of Calzlfornia, Medical Ceder , San Fral~cisco, Ca,lif. 94143, U.S.A. (Received 81 I\‘eptemhw 1978, New York) The angiogenic activities of the fibroblast and epidermal growth factors in vivo have been analyzed using the rabbit cornea as a model. The angiogenic activity of slow-release form of Elwax 40 containing either fibroblast growth factor (FGF) or epidermalgrowth factor (EGF) at concentrations ranging from 0.5 to 6Opg were compared. A positive responsewas observed with concentrations of EGF as low as 1 pg per implant. The optimal concentration causing neovasrularization in lOO?& 0 f the cases was 1Opg per implant. With FGF, a posit,ive response was observed at 2.5 pg per implant and the optimal concentration causing neovascularization in lOO’-‘~ of the cases was 25~g per implant. In both cases, and in contrast with implants of serum albumin, the neovascularization could be seen to progress in the absence of inflammation. Key ~orr1.r : angiogenesis : cornea ; EGF ; FGF.
1. Introduction Fibrohlast growth factor (FGF) has been shown to be a mitogen for a wide variety of cells derived from the mesoderm and maintained in tissue culture (Gospodarowjcz, 1976; Gospodarowicz, Mescher and Birdwell, 1978; Gospodarowicz, Vlodavsky, Fielding and Birdwell, 1978). A mong the different cell types tested, bovine or human vascular endothelial cells (Gospodarowicz, Moran, Braun and Birdwell, 1976; Gospodarowicz, Moran and Braun, 1977; Gospodarowicz, Brown, Birdwell and Zetter, 1978), coming from vascular tissue derived from either fetal or adult individuals (umbilical vein, the aortic arch, and the endocardium) have been shown to respond particularly well to the mitogenic effect of FGF. Although most of the studies performed on the mitogenic effect of FGF with vascular endothelial cells has ljeen done using cells maintained in tissue culture, there is the distinct possibility that FGF could also be mitogenic for the vascular endothelium in vivo. Two models have been developed for in vivo studies of vascular endothelial cell proliferation. The first model consists in following the rate of regeneration of the endothelium after denuding the intima of a vein or an artery. This model is best adapted to the study of mitogens acting on the vascular endothelium of big vessels. However, due to the relative inaccessibility of these anatomical sites, continuous observation of the progress of the endothelial regeneration reflecting migration and proliferation is difficult. The second model takes advantage of the observation that neovascularization within an avascular space such as the cornea results from the rapid proliferation of vascular endothelial cells which first form solid cords and later organize into new capillaries. This model is best adapted to the study of the proliferation of the capillary endothelium. By losing the cornea as a model, therefore. one can easily study the effect of angiogenic agents on the proliferation of invading capillaries which results directly from an increased proliferation of vascular endothelial cells (Gimbrone, Leapman and 0014-4535/79/050501~14 $01.00/0 A
0 1979 Academic Press Inc. (London) Limited 501
508
I). 00SI’0DAROW’ICZ.
H. RIALEPKI
AND
T. K. THAKRAL
Folkman. 1974). Such studies have been made even easier by the recent development of a slow-release form polymer such as Elwax 40, a copolymer made of Ethylenevinyl acetate (Langer, Brem, Falterman, Klein and Folkman, 1976) or that’ of Hydron S, a polymer made of hydroxyethylmethacrylate (Langer and Folkuran. 1976). These two slow release agents have been shown to be non-inflammatory anal to he capable of releasing various proteins for long periods of time (Langer, Brtm et al., 19’76; Langer and Folkman, 1976).
2. Materials and Methods Materials
FGF was purified, as previously described, from bovine pituitary glands (Gospod:browicz, 1975; Gospodarowicz, Bialecki and Greenburg, 1978). Pituitary FGF yielded a single bantl in polyacrylamide disc gel electrophoresis at pH 4.5. EGF was purified as described bv Savage and Cohen (1972) from the sub-maxillary glands of adult male Swiss Webster mice that had been given subcutaneous injections of testosterone propionate (1 nlg per animal) daily for 8 days. The final preparation yielded a single band on polyacrylamide gel electrophoresis at pH 8.5, and three amino acids--lysine, alanine, and phenylalanit~e were t:bsent from the fIna preparation as is characteristic of EGF (Savage and Cohen, 1974). The biological activity of EGF, as measured by the stimulation of the initiation of D?iA synthesis in human foreskin fibroblasts, was indistinguishable from that of a reference preparation sent by Dr S. Cohen (Vanderbilt I’niversit)y, Nashville, Tenn.). Elwax 40 (Ethylenevinyl acetate copolymer) was a generous gift from Dr Wong (Alza Corp., Palo Alt,o, Cn.). Bovine serum albumin crysCallizer1 three times (BSA) was obtained from Schwarz Mann Co. (Orangeburg, N.Y.).
(a) Preparatios of the Elwuz 40 slowrelease form polymer of EGF, FGF and B8d Elwas 40, a copolymer of Ethylenevinyl acetate first washed extensively with ethanol, was dissolved in methylene chloride (CH,Cl,) at a final concentration of 120/, (W/V). As described by Langer and Folkman (1976), a drop of the solution (100~1 covering a surface area of 25 mmz) way applied on a glass slide and left to dry under :I slight vacuum for 2 hr. After a thin and transparent, film of copolymer had formed, 100 ~1 of a la”;, Elwax solution in CH,Cl, containing in suspension either 10, 50, 100, 250. 500 pp or 1.5 mg of either EGF or FGF and 1 or 10 mg of BSA were then added on top of the film and dried up as already described. The thin and transparent film of Elwax 40 containing either EGF, FGF or BSA, which appeared as a granular material, was then cut in squares (1 mm2) with a razor blade. Each implant of FGF or EGF slow release form polymer cont,ained respectively 0.5, 1, 2*5,5,10,20 and 60 pg of proteins, while the BSA slow release form polymer contained either 40 or 4( I() pg of proteins. (b) Zrrlplantatiorl technique. The implantation technique way similar to that described by Gimbrone, Cotran, Leapman and Folkman (1974). Th e rabbits were anaesthetized with itu intravenous injection of pentobarbital (25 “g/kg). The eye was moved forward alld secured in posibion by a fold clamped in the lower lid. With a Bard-Parker #ll blade a superficial incision 1.5 mm long was made in the cornea1 dome to one side of its cellter. The incision was then continued down into, but not t)hrough, the cornea. A malleable iris spatula (1.5 mm width) was inserted, and an oblong pocket fashioned within the cornea1 strama. Peripheral pcckets ended 2-3 mm from the limbus (cornea sclera junction). The implant (1 mm2), which consisted either of the slow-release form of Elwax 40 alone or of a slow-release form of Elwas 40 containing EGF, FGF or BSA, was then deposited in the bottom of each pocket, which then sealed spontaneously.
FIBROBLAST
AND
EPIDERMAL
GROWTH
FACTOR
503
Strrrorr~icrosco~~ic observations Eyes wit#h cornea1 implants were examined daily by two observers through a Zeiss Slit Lamp Stereomicroscope (Carl Zeiss, Inc.) at X10-X40. Implant and new vessel growth were measured en face with an ocular micrometer at 10 5. Empty cornea1 pockets fashioned at distances 1 mm from the limbus never stimulated cornea1 neovascularixation. The eyes were photographed every other day with a Canon camera equipped with a close-up lells.
After 7 to 14 days, entire eyes were formalin. The eyes were then embedded lin and eosin. Photomicrographs were bright field objectives and an aut,omatic Bvaluntiori
excised and fixed by immersion in 10~~ buffered in paraffin, sectioned, and stained with hematoxytaken with a Nikon microscope equipped with camera.
qf the extent of the neol~asclllarizat~on
of the corseal
stroma
A positive neovascular response wva,sobserved when capillaries in the shape of hairpin loops (scored as a weak response) or sprouting as a dense brushwork (scored as a strong respollse) grew from the adjacent limbal plexus in a directional wav toward the implant. Corneas wit,h a positive response stayed clear for the length of the experiment and did not develop an edema. x’o increased cornea1 thickness, as seen by slit lamp measurement, c~~ultl be observed between days 7 and 14. Histologically, few or no macrophages were found in the cornea1 stroma and no inflammatory cells could be detected 7 to 14 days irfter inlplantation. Responses were scored as negative when no growth of capillaries was observed or when a few capillaries in the shape of a hairpin loop developed during the first few days but) later regressed. When more than 102 polymorphonuclear leukocytes was attributed to the were detectecl per histological section, the neovascularization i tlflammati(lu procrsn rather than t,o an effect of growth factor.
EGF and FGF were suspended in CH,CI, at a final concentration of 1 rng per IOO pl. The suspension was dried under vacuum overnight. The dry powder was then redissolved in 1 ml DME containing @Sqb crystallin BSA and filtered through a 0.2 p Jlillipore filter to sterilize the solution. After diluting t,he EGF and FGF solutions with (j.57; BSA in l)?IJE at appropriate concentrations (50 &ml), the ability of the two mitogens to stimulate DNA synthe,+ in 3T3 cells maintained in low serum concentration was measured as alreadv desc,ribed (Gospodarowicz, Bialecki and Greenburg, 1978) and compared to that of rut& EGF am1 FGF.
3. Thr eflect of
CH,Cl,
treatment
Results
OPLthe mitogewic
properties
of
EGF nn,d FGF
Bince it is known that exposure of proteins to organic solvents can have a deleterious effect on their biological properties: we have tested the effect of CH,CI, treatment on the mitogenic properties of both EGF and FGF. EGP or FGF were first suspended in C’H,CI, at a final concentration of 10 mg/ml and then dried. When their effects on the initiation of DNA synthesis in resting populations of Balb/c 3T3 was compared to those of native EGF or FGF, they were found to be identical (Fig. 1). It was therefore demonstrated that the treatment with (‘H&I, of either FGF or EGF does not affect their mitogenic properties.
504
D.
GOBPODAROWICZ,
H.
BIALEC’KI
6
AND
T.
FGF
t
K.
THAKRAL
FGF -
100 ng
IO ng
F‘rc:. 1. DR’A synthesis in ST3 cells induced by pitnit.ary PCX or EC:P before and after exposure of the mitogens to methylem= chloride. The cells were plated and maintained as described by Gospodarouicz, Greenburg and I3ialecki (IQTY). The incorporation of [3H]t,hymidine into DNA was dct~ermined as described by Gospodarowicz, ($rcenburg and Bislocki (1978). Both mitogens were tested at roncentrations of 1,lO and 100 x$ of proteins/ml. Native FGF (W), FGF exposed to methyleno chloride ( .--), native EGF (m), EGF exposer1 to methylene chIorido ( . ‘), control (c).
The mragiogewic eflect of slow-released EGP
When slow-release form implants of EGP ranging in concentrations from O-5 to 60 pg of EGF were implanted in the cornea1 stroma: a positive response (neovascularization of the cornea1 stroma) was observed with concentrations ranging from 1 to 60 pg per implant. No response was elicited by implants containing 0.5 pg of EGF (Table I). TABLE
Comeal vascularizntion
Number
of corma tested
EGF
1
induced by CCslow released fowl
Seovsscularization Positive Xegative
concentration
60M 20 PLg 1OM 5PLg 2.5 pg l/Lg 0.5 fF
Inflsmmation Positive Negative
.._____
-__ 5 5 6 8 8 5 5
of EGE
5 6 6 5 5 2 1
0 0 0 2 3 3 4
~~ 0 0 1 1 2 0 0
5 5” 6 7 5 5
FIG. 2. Proliferation of capillaries into the rabbit cornea induced by au Elwax 40 implant. An Elwax 40 implant containing either 5pg of EGF (a), 1Opg of FGF (b), 4Opg of BSA (c), or nothing (d), was implanted into the cornea1 stroma (arrows). In the case of the Elwax implant alone or containing BSA, no detectable capillary proliferation takes place. In the case of the implant containing either FGF or EGF, capillaries proliferate from the limhus t,oward the implant.
506
D. GOSPODAROWICZ,
H. BIALECKI
ASD
T. K. THAKRAL
With 1 pg of EGF, a positive response was olbserved in 40:;) of the cases, while with 2.5 and 5 pg it was observed in 60 and 70”; of the cases respectively. In all positive cases. capillaries started to invade the cornea1 stroma 3 -4 days following iml~lant~ation and reached the implant within 8 10 (lays. invading 2-3 mni of cornea1 stroma. In all cases, t,he capillaries had the grc)ss morphological appearance of an elongated hairpin progressing toward the implant. No capillaries were ohservccl within the corneai stroma in regions far away from the implant [Fig. 2(a)]. In 110 case was inflammation detectable, as reflected by the lack of cornea1 edema or the absence of an hypopyon in the anterior chamber. The cornea stayed clear at all times. With concentrations of EGF ranging from 10 to 60 pg per implant, a positive response was observed in lOOy, of the cases, Capillaries invading the cornea1 stroma started to appear as early as two days after implantation and developed into a brush border which reached the implant within five clays. They afterwards developed into 2 or 3 main Lvascular trunks which regressed 4-G weeks later, probably because of the depletion of EGF in the implant. The histological appearance of corneas implantecl
FIG. 3. Histological examination 14 days post implantation of the efi’ect of an Elwax 40 implant containing 10 CL::of EGF on the structure of the cornea. (a) ~lomoal pocket. The epithelium over thts cornea1 pocket shows clear signs of hyperplasia ( :: 40). (b) ( ‘entral part of the cornea. No cpithelial hyperplnsia is ohserved (x40). (c) Same as (a) but showing the cornea1 epithelium (X 100). (d) &NW as (b) but showing the cornea1 epithelium ( :< 100). (a) D i ff erentiated blood vessels ( ~200). (f) Blood vessels infiltrating the cornea1 stroma (x 100).
FIBROBLSST
AND
EPIDERMAL
GROWTH
PAC’TOR
50s
Fro. 4. Presence of polymorphonuclear eosinophilic leukocytes in the stroma of cornea carrying an Elwas 40 implant containing either 5 pg of EGF (a) ( ‘I\ 100) or 40 pg of BYA (b) ( x 100). In both cases. nrovascularization of the cornea1 stroma developed after a period of 3 days. While with the EGF implant few or no inflammatory cells could be seen, (only eosinophylic leukocytes, could be detected (a) arrows, : 400). In the case of an Elwax 40 implant containing BSA, there was a massive infiltration of t,h+t stroma by polymorphonuclear leukocytes (b), ( .< 100).
508
D. GOSPODAROWICZ,
H. BIALECKI
AND
T. K. THAKRAL
by slow-release form of EGF is shown in Fig. 3. The cornea1 epithelium located over the cornea1 pocket containing the implant, was distinctly thicker than that located far away from the implant [Fig. 3(a), (b)]. While the cornea1 epithelium consisted of a basal cell layer and three layers of wing cells [Fig. 3(d)] in the part of the cornea remote from the implant, in the region overlying the implant as many as 16 layers of wing cells could be seen on top of the basal cell layer [Fig. 3(c)]. The morphological appearance of capillaries growing into the stroma is shown in Fig. 3(e), (f). The absence of inflammatory cells such as mononuclear cells and pseudoeosinophilic leukocytes, which are conspicuous in the rabbit because of their tinctorial properties, is evident. Only in rare cases were inflammatory cells present, and when present [Fig. 4(a)] only a few (less than 5) could be detected in each section (Table I, II: III). In some cases: inflammation could be detected, as was reflected by the presence of macrophages and other inflammatory cells: but tjhe total number of these cells was never higher than lo6 per whole cornea, as was determined hy counting inflammatory cells in cornea1 sections [Fig. 4(b)]. These corneas were recorded as inflamed (Table I) and were not scored as positive for either FGF or EGF. TABLE
Number of cornea tested
10 12
The angiogenic
PC’F T concentration
400 ‘40
II.
Neovasculerization Positive Kegstive
5 4
3 6
Inflammation Positive Ntyative
7 1
3 6
e#ect of slozu release form of FGF
When slow-release form of FGF ranging in concentrations from 0.5 to 6Op.g was implanted in the cornea1 stroma, a positive response was observed with concentrations ranging from 2.5 to 6Opg (Table II). No response was observed at a concentration of 0+5pg. The response observed with an implant containing 1 pg of FGF was in one case attributed to a local inflammation (Table II). With 2.5p.g of FGF a positive response was observed in SO:/, of the cases, while
FIBROBLAST
AKD
EPIDERMBL
GROWTH
FACTOR
509
with 5 pg of FGF it was observed in 85% of the cases. With either concentration, the neovascularization can be considered to be a weak response, since a response was not seen before day 4 and reached the implant only 4-8 days later. As already observed with low concentrations of EGF, the capillaries had the gross morphological appearance of an elongated hairpin. In one of 44 cases, a pronounced inflammation was observed. Histological examination of the corneas revealed the presence of macrophages and inflammatory cells in only 3 of 44 cases. These corneas were scored as presenting signs of inflammation (Table II). Wit,h concentrations of FGF ranging from lo-60 pg in the implantation, a posit’ive response was achieved in 100% of the cases, and, as with high EGF concentrations, two days follow-ing the implantation the corneas exhibited capillaries growing into the cornea1 stroma as a brush border and reaching the implant 4 days later [Fig. 2(b)]. ,2s wit’li the EGF implant, in rabbits which were not sacrificed, the capillaries began to regress 4-6 weeks later.
FIG. 5. Histological examination, 14 days post implantation, of the effect of an Elwax 40 implant containing 10 pg of FGF on the structure of the cornea. (a) Cornea1 pocket, partly jelled with red blood cells. (b) Region of the cornea adjacent to the limbus. Blood vessels can be seen growing in the upper part of the cornea1 stroma (arrows) (x20). (c) Same as (a), but showing the cornea1 pocket at a higher magnification (;i 100). (d) Same as (b), but showing the upper part of the cornea1 stroma where capillaries are present at a higher magnification ( x 100). (e) Capillaries and cornea1 stroma ( x 200). (f) A group of capillaries and blood vessels ( x 400).
510
D. GOSPODAROWICZ,
H. BIALECKI
ASD
T. K. THAKRAL
The histological studies of corneas with implants containing FGF are shown in Fig. 5. In contrast to corneas implanted with EGF in Elwsx 40, there \vas no epithelial hyperplasia of the cornea1 epithelium underlying the cornea1 stroma. nor was there any significant number of eosinophil leukocytes or mononuclear cells. In some cases, the cornea1 pocket which had no time to completely seal co~~ltl he seen to fill witjh red blood cells when vessels reached it.
When the angiogenie effect of I&ax 40 alone s-as tested in the cornea, it was found to be inactive. In 10 cases out of 10. it, did not induce any vascularization of t,he cornea (Table III). Histological examination of the corneas show that. the Elwax 40 implant~s did not induce any epithelial hvperplasia nor did they seem to have an! significant, effect on the kemtocytes of the cornea1 stroma [Fig. 6(a). (b), (c)].
Fx. 6. IIistologicsl ennminetion, 14 days post implantstion, of the effect of an E1w1.u 40 implsut on the structure of the cornea. (a) Corned pocket ( \: 20). (b) Region of the cornea betwcen the corncal pocket and the limbus. No blood vessels arc prcsrnt ( x 20). (c) Corned pocket seen at, u higher magnification (x 100).
When Elwax implants containing BSA were implanted in the stroma, either 40 or 400 pg of protein failed to induce vascularization in six cases out of 10. and three cases out of 10 respectively (Table III). In the four cases (40 yg BSA) and seven cases (400 pg B&4) where vascularization was observed, there was a massive infiltration of leukocytes (mononuclear cells) into the cornea1 stroma. Numerous inflammatory cells (eosinophylic leukocytes) and macrophages could be observed. 4. Discussion Capillary proliferation has been shown to be a general feature of actively growing tissue, whether they are of normal origin, such as the corpus luteum (Gospodarowicz and Thakral, 1978), salivary gland (Hoffman, McAushan, Robertson and Burnett, 1976), and granulation tissue (Algire and Chalkley, 1965), or from tumors, such as
FLHRORLAST
APJD EPIDEKMSL
GROWTH
FACTOR
511
solid malignant tumors represented by the Brown Pearce epithelioma, the V, carcinoma, B16 mouse melanoma, and by the sarcoma 37, (Folkman, 1974; Folkman and Cotran, 1976). Neovascularization can also be induced by extracts from various cell types derived from either tumors (Phillips, Steward and Kumar, 1976), established cell lines such as Balb/c 3T3: (Klagsburn, Knight,on and Folkman, 1976). or from neutrophils (Fromer and Klentworth, 1976). It has been demonstrated that’ at. least in the case of tumor tissue one can purify an angiogenic factor (TSF) which. when implanted into the cornea in the form of a slow-release form of a pol,vmcr consisting of an acrylamide pellet, induces neovascularization in vivo (Folkman. !Merler, Abernathy and Williams. 1974; Phillips, St’eward and Kumar, 1976). No studies on the angiogenic potency in vivo of mitogens such as the epidcrrnal or fibroblast growth factors have been conducted. although there is a dist’inct possiIjility t(hat these factors could be angiogenic in viva. For example, FGF has ljeen shown to l)e a strong mitogen for cultured vascular endothelial cells. EGF is present, at high concentrations in mouse submaxillary glands and crude extracts from this gland are angiogenic in vivo (Folkman and C’otran. 1976). In addition, Savage and ( ‘ohen reported in 1972 that the topical administration of EGF to corneas denutletl of their epit’helium resulted not only in a rapid reepithelization of the denuded area but also induced the blood vessels in the limhal region to become much more promin ent in t’he EGF-treated eye than in the control. It is conceivable? therefore, that although EGF does not have a pronounced mitogenic eflect on vascular endothelial cells in vitro. it’ could have an effect, on neovascularization in viva, either directSly OI indirectly. When the anpiogenic properties of both EGF and FGF are testecl in viva, using for clelivcry a slowrelease form polymer composed of Elmax 40 implanted in the rabtnt cornea (a tissue normally transparent and avascular), our results indicate that’ t)oth mitogenic agents induce capillary proliferation. The optimal concentrat,ion of mitoprns required for such an effect is in the range of 10 pg. When tested at sul)optimxl concentrations, EGF is a more potent angiogenic agent than FGF. Although the concentrations of mitogens used in these studies are 100. to lOOO-fokl higher than bhose required in vitro t,o see a mitogenic effect on nlesotlerm-derivetl cells, one has to keep in mind that it is unlikely, when one uses a slow-release forni p(Jlpler such as Elwax 40, that all the mitogen will be delivered at once at the limbal region. ;Iccording to the data of Langer and Folkman (1976), after the mibogcnic agent is released from the polymer in an initial burst, it is later released at minutc~ concentrations but at a sustained rate. and this process can last for weeks. The concentrahion of mitogen which will diffuse through the cornea1 stroma on a pr diew basis will in all likelihood be much smaller than the original amount, put into the polynicr. One of the pit~falls in the study of neovascularization in the cornea has been the dilhcultv of distinguishing between the straight angiogenic effect of a mitopenic agent and the induced inflammatorv process caused by tissue extracts or organic agents (polymer) inserted into the cornea1 stroma. This inflammatory process has been recognized in all cases to be the most potent inducer of angiogenesis and therefore represents a potentially troublesome source of artefact in studies of cornea1 capillary proliferation. Although it has been considered by some that the clarity of the cornea, the lack of cornea1 edema. and the lack of hypopyon in the anterior chamber could reflect a lack of inflammation. these are only signs which may also occur in t’he acute phase of the inflammation process. It has been shown by others that, although a cornea can stay perfectly clear, an invasion by polymorphonuclear
512
D. QOSPODAROWIC‘Z,
H. BIALECKI
AND
T. K. THAKRAL
cells can precede a phase of vascularization (Fromer and Klintu-orth, 1976). It has also been shown that activated macrophages implanted into the cornea1 stroma are a source of factor(s) Rhich can induce vascularization (Clark, Stoney, Leung, silver, Hohn and Hunt, 1976; Polverine, Cotran, Gimbrone and Unanue, 1977). For these reasolls, therefore, a study of tbe angiogenic potential of mitogens implanted in the cornea1 stroma should be accompanied by a histological study of the cell types present during the active phase of capillary proliferat.ion. Although in this study the presence of inflammatory cells was not examined during the initial phase of capillary proliferation (first. seven days) but only 7-14 days after implantation. our results demonstrate that. in the case of corneas containing either an FGF or EGF implant, an invasion by polymorphonuclear leukocytes and nlacrcJphages is unlikely to be the cause of the long-term capillary proliferation. In all cases. these cell types were either ahsent from the cornea1 stroma, particularly around the cornea1 pocket where they should accumulate, since it is the densest region of capillaries, or, when present, their total concentration was far t)clow that used tjv others to induce neovascularizat,ion. In the case of inflammation, for esample, a massive invasion of t,he whole cornea1 stroma by polymorphonuc1ea.r leukocytes has been reported by Framer and Klintworth (1976)> while in the case of luaorophages, no less than 5 x 1Oj cells are required for an angiogenic response (Clark et al.. 1976; Polverine et a,l., 1977). That t’he angiogenic properties of FGF or EGF are distinct from those of leukocyte- or macrophage-mediated events can be further deducecl from the observation that, when BSA is llsed, in every case where one ohserves vascularization, it correlates with a massive invasion of the cornea1 strorna 1)~ leukocytes. No such correlat,ion could be observed when t,he neovascularization was induced hv e&her FGF or EGF. This observation tends to demonst.rate t.hat two diflerent types of neovascularization can be observed in vivo, one of which is induced hp a specific and purified mitogen acting at low concentrations. This type of vascularization is not mediated by leukocytes or rnacrophage,,res. In contrast. the neovascularizat,ion induced by non-specific agent)s, such as BSA. which has no nlitogcnic a&iv+per se but is capable of inducing an infiammation. could \JC niediatctl 1)~ leukocyt~cs anal macrophages. Finally, one has to resolve whether the neovAscLllarizati(jII inclucetl in viva I)y either FGF or EGF ha,s a direct or indirect effect on the vascular endot~helium. In the case of FGF: which has been shown to be mitogenic for vascular entlothelial cells maintained in vitro, the possibility exists that. in vivo it could directly stimulate the proliferation of capillary endothelial cells. With EGF‘, such a direct effect is 1)~ no mea,ns evident. First, EGF has been shown to be either inactive or poorly rnitogenic for bovine va,scular endothelial cells and human endothelial cells, respectively. Therefore, there is no reason why it should act directly on the capillary endothelium in vivo. Secondly, EGF has been shown to be a potent mitogen for the cornea1 epithelium, which responds to it with a striking hyperplasia. It is therefore possible that the neovascuIarization observed in the case of EGF is not a direct effect of the mitogen but is secondary to the cornea1 epithelial hyperplasia. Alternatively, EGF could directly trigger the proliferation of capillary endothelial cells: although it had little effect on this cell type in vitro.
FIBROBLAST
AND
EPIDERMAL
GROWTH
FACTOR
513
ACKNOWLEDGMENTS
This work was supported by grants from the NH (EY 01068 and NICHD-11082) to Dr Gospodarowicz and (GM-12828) to Dr T. Hunt. W e wish to thank Mr Harvey Scodel for his invaluable assistance in the preparat,ion of this manuscript.
REFERENCES Algire, G. H. and Chalkley, H. W. (1945). Vascular reactions of normal and malignant tissues in vivo I. Vascular reactions of mice to wounds and t,o normal and neoplastic transplants. Nufl. Canc,er Inst. 6, 75-85. Clark, R. A., Stoney, R. D.. Leung, D. Y. K., Silver, I., Hohn, D. C. and Hunt, T. K. (1976). Role of macrophages in wound healing. Surg. Forum 27, 16-19. Folkman, J. (1974). Tumor angiogenic factor. Cnncer Res. 34, 2109-23. Folkman, J. and Cotran, R. (1976). Relation of capillary proliferation to tumor growth, In RP~‘. Exptl. Pathol. (Ed. Richter, G. W.). Pp. 207-48. Fromer, C. H. and Klintworth, G. I
Eq.
The Angiogenic Activity of the Fibroblast and Epidermal Growth Factor D. GOSPODAROWICZ*~, H. BIALECKI*
AND T. K. THAKRAL,!:
*Cancer Research Institute, Departm,ent of tMedicine wad @Twrgery, lrGvers7Ity of Calzlfornia, Medical Ceder , San Fral~cisco, Ca,lif. 94143, U.S.A. (Received 81 I\‘eptemhw 1978, New York) The angiogenic activities of the fibroblast and epidermal growth factors in vivo have been analyzed using the rabbit cornea as a model. The angiogenic activity of slow-release form of Elwax 40 containing either fibroblast growth factor (FGF) or epidermalgrowth factor (EGF) at concentrations ranging from 0.5 to 6Opg were compared. A positive responsewas observed with concentrations of EGF as low as 1 pg per implant. The optimal concentration causing neovasrularization in lOO?& 0 f the cases was 1Opg per implant. With FGF, a posit,ive response was observed at 2.5 pg per implant and the optimal concentration causing neovascularization in lOO’-‘~ of the cases was 25~g per implant. In both cases, and in contrast with implants of serum albumin, the neovascularization could be seen to progress in the absence of inflammation. Key ~orr1.r : angiogenesis : cornea ; EGF ; FGF.
1. Introduction Fibrohlast growth factor (FGF) has been shown to be a mitogen for a wide variety of cells derived from the mesoderm and maintained in tissue culture (Gospodarowjcz, 1976; Gospodarowicz, Mescher and Birdwell, 1978; Gospodarowicz, Vlodavsky, Fielding and Birdwell, 1978). A mong the different cell types tested, bovine or human vascular endothelial cells (Gospodarowicz, Moran, Braun and Birdwell, 1976; Gospodarowicz, Moran and Braun, 1977; Gospodarowicz, Brown, Birdwell and Zetter, 1978), coming from vascular tissue derived from either fetal or adult individuals (umbilical vein, the aortic arch, and the endocardium) have been shown to respond particularly well to the mitogenic effect of FGF. Although most of the studies performed on the mitogenic effect of FGF with vascular endothelial cells has ljeen done using cells maintained in tissue culture, there is the distinct possibility that FGF could also be mitogenic for the vascular endothelium in vivo. Two models have been developed for in vivo studies of vascular endothelial cell proliferation. The first model consists in following the rate of regeneration of the endothelium after denuding the intima of a vein or an artery. This model is best adapted to the study of mitogens acting on the vascular endothelium of big vessels. However, due to the relative inaccessibility of these anatomical sites, continuous observation of the progress of the endothelial regeneration reflecting migration and proliferation is difficult. The second model takes advantage of the observation that neovascularization within an avascular space such as the cornea results from the rapid proliferation of vascular endothelial cells which first form solid cords and later organize into new capillaries. This model is best adapted to the study of the proliferation of the capillary endothelium. By losing the cornea as a model, therefore. one can easily study the effect of angiogenic agents on the proliferation of invading capillaries which results directly from an increased proliferation of vascular endothelial cells (Gimbrone, Leapman and 0014-4535/79/050501~14 $01.00/0 A
0 1979 Academic Press Inc. (London) Limited 501
508
I). 00SI’0DAROW’ICZ.
H. RIALEPKI
AND
T. K. THAKRAL
Folkman. 1974). Such studies have been made even easier by the recent development of a slow-release form polymer such as Elwax 40, a copolymer made of Ethylenevinyl acetate (Langer, Brem, Falterman, Klein and Folkman, 1976) or that’ of Hydron S, a polymer made of hydroxyethylmethacrylate (Langer and Folkuran. 1976). These two slow release agents have been shown to be non-inflammatory anal to he capable of releasing various proteins for long periods of time (Langer, Brtm et al., 19’76; Langer and Folkman, 1976).
2. Materials and Methods Materials
FGF was purified, as previously described, from bovine pituitary glands (Gospod:browicz, 1975; Gospodarowicz, Bialecki and Greenburg, 1978). Pituitary FGF yielded a single bantl in polyacrylamide disc gel electrophoresis at pH 4.5. EGF was purified as described bv Savage and Cohen (1972) from the sub-maxillary glands of adult male Swiss Webster mice that had been given subcutaneous injections of testosterone propionate (1 nlg per animal) daily for 8 days. The final preparation yielded a single band on polyacrylamide gel electrophoresis at pH 8.5, and three amino acids--lysine, alanine, and phenylalanit~e were t:bsent from the fIna preparation as is characteristic of EGF (Savage and Cohen, 1974). The biological activity of EGF, as measured by the stimulation of the initiation of D?iA synthesis in human foreskin fibroblasts, was indistinguishable from that of a reference preparation sent by Dr S. Cohen (Vanderbilt I’niversit)y, Nashville, Tenn.). Elwax 40 (Ethylenevinyl acetate copolymer) was a generous gift from Dr Wong (Alza Corp., Palo Alt,o, Cn.). Bovine serum albumin crysCallizer1 three times (BSA) was obtained from Schwarz Mann Co. (Orangeburg, N.Y.).
(a) Preparatios of the Elwuz 40 slowrelease form polymer of EGF, FGF and B8d Elwas 40, a copolymer of Ethylenevinyl acetate first washed extensively with ethanol, was dissolved in methylene chloride (CH,Cl,) at a final concentration of 120/, (W/V). As described by Langer and Folkman (1976), a drop of the solution (100~1 covering a surface area of 25 mmz) way applied on a glass slide and left to dry under :I slight vacuum for 2 hr. After a thin and transparent, film of copolymer had formed, 100 ~1 of a la”;, Elwax solution in CH,Cl, containing in suspension either 10, 50, 100, 250. 500 pp or 1.5 mg of either EGF or FGF and 1 or 10 mg of BSA were then added on top of the film and dried up as already described. The thin and transparent film of Elwax 40 containing either EGF, FGF or BSA, which appeared as a granular material, was then cut in squares (1 mm2) with a razor blade. Each implant of FGF or EGF slow release form polymer cont,ained respectively 0.5, 1, 2*5,5,10,20 and 60 pg of proteins, while the BSA slow release form polymer contained either 40 or 4( I() pg of proteins. (b) Zrrlplantatiorl technique. The implantation technique way similar to that described by Gimbrone, Cotran, Leapman and Folkman (1974). Th e rabbits were anaesthetized with itu intravenous injection of pentobarbital (25 “g/kg). The eye was moved forward alld secured in posibion by a fold clamped in the lower lid. With a Bard-Parker #ll blade a superficial incision 1.5 mm long was made in the cornea1 dome to one side of its cellter. The incision was then continued down into, but not t)hrough, the cornea. A malleable iris spatula (1.5 mm width) was inserted, and an oblong pocket fashioned within the cornea1 strama. Peripheral pcckets ended 2-3 mm from the limbus (cornea sclera junction). The implant (1 mm2), which consisted either of the slow-release form of Elwax 40 alone or of a slow-release form of Elwas 40 containing EGF, FGF or BSA, was then deposited in the bottom of each pocket, which then sealed spontaneously.
FIBROBLAST
AND
EPIDERMAL
GROWTH
FACTOR
503
Strrrorr~icrosco~~ic observations Eyes wit#h cornea1 implants were examined daily by two observers through a Zeiss Slit Lamp Stereomicroscope (Carl Zeiss, Inc.) at X10-X40. Implant and new vessel growth were measured en face with an ocular micrometer at 10 5. Empty cornea1 pockets fashioned at distances 1 mm from the limbus never stimulated cornea1 neovascularixation. The eyes were photographed every other day with a Canon camera equipped with a close-up lells.
After 7 to 14 days, entire eyes were formalin. The eyes were then embedded lin and eosin. Photomicrographs were bright field objectives and an aut,omatic Bvaluntiori
excised and fixed by immersion in 10~~ buffered in paraffin, sectioned, and stained with hematoxytaken with a Nikon microscope equipped with camera.
qf the extent of the neol~asclllarizat~on
of the corseal
stroma
A positive neovascular response wva,sobserved when capillaries in the shape of hairpin loops (scored as a weak response) or sprouting as a dense brushwork (scored as a strong respollse) grew from the adjacent limbal plexus in a directional wav toward the implant. Corneas wit,h a positive response stayed clear for the length of the experiment and did not develop an edema. x’o increased cornea1 thickness, as seen by slit lamp measurement, c~~ultl be observed between days 7 and 14. Histologically, few or no macrophages were found in the cornea1 stroma and no inflammatory cells could be detected 7 to 14 days irfter inlplantation. Responses were scored as negative when no growth of capillaries was observed or when a few capillaries in the shape of a hairpin loop developed during the first few days but) later regressed. When more than 102 polymorphonuclear leukocytes was attributed to the were detectecl per histological section, the neovascularization i tlflammati(lu procrsn rather than t,o an effect of growth factor.
EGF and FGF were suspended in CH,CI, at a final concentration of 1 rng per IOO pl. The suspension was dried under vacuum overnight. The dry powder was then redissolved in 1 ml DME containing @Sqb crystallin BSA and filtered through a 0.2 p Jlillipore filter to sterilize the solution. After diluting t,he EGF and FGF solutions with (j.57; BSA in l)?IJE at appropriate concentrations (50 &ml), the ability of the two mitogens to stimulate DNA synthe,+ in 3T3 cells maintained in low serum concentration was measured as alreadv desc,ribed (Gospodarowicz, Bialecki and Greenburg, 1978) and compared to that of rut& EGF am1 FGF.
3. Thr eflect of
CH,Cl,
treatment
Results
OPLthe mitogewic
properties
of
EGF nn,d FGF
Bince it is known that exposure of proteins to organic solvents can have a deleterious effect on their biological properties: we have tested the effect of CH,CI, treatment on the mitogenic properties of both EGF and FGF. EGP or FGF were first suspended in C’H,CI, at a final concentration of 10 mg/ml and then dried. When their effects on the initiation of DNA synthesis in resting populations of Balb/c 3T3 was compared to those of native EGF or FGF, they were found to be identical (Fig. 1). It was therefore demonstrated that the treatment with (‘H&I, of either FGF or EGF does not affect their mitogenic properties.
504
D.
GOBPODAROWICZ,
H.
BIALEC’KI
6
AND
T.
FGF
t
K.
THAKRAL
FGF -
100 ng
IO ng
F‘rc:. 1. DR’A synthesis in ST3 cells induced by pitnit.ary PCX or EC:P before and after exposure of the mitogens to methylem= chloride. The cells were plated and maintained as described by Gospodarouicz, Greenburg and I3ialecki (IQTY). The incorporation of [3H]t,hymidine into DNA was dct~ermined as described by Gospodarowicz, ($rcenburg and Bislocki (1978). Both mitogens were tested at roncentrations of 1,lO and 100 x$ of proteins/ml. Native FGF (W), FGF exposed to methyleno chloride ( .--), native EGF (m), EGF exposer1 to methylene chIorido ( . ‘), control (c).
The mragiogewic eflect of slow-released EGP
When slow-release form implants of EGP ranging in concentrations from O-5 to 60 pg of EGF were implanted in the cornea1 stroma: a positive response (neovascularization of the cornea1 stroma) was observed with concentrations ranging from 1 to 60 pg per implant. No response was elicited by implants containing 0.5 pg of EGF (Table I). TABLE
Comeal vascularizntion
Number
of corma tested
EGF
1
induced by CCslow released fowl
Seovsscularization Positive Xegative
concentration
60M 20 PLg 1OM 5PLg 2.5 pg l/Lg 0.5 fF
Inflsmmation Positive Negative
.._____
-__ 5 5 6 8 8 5 5
of EGE
5 6 6 5 5 2 1
0 0 0 2 3 3 4
~~ 0 0 1 1 2 0 0
5 5” 6 7 5 5
FIG. 2. Proliferation of capillaries into the rabbit cornea induced by au Elwax 40 implant. An Elwax 40 implant containing either 5pg of EGF (a), 1Opg of FGF (b), 4Opg of BSA (c), or nothing (d), was implanted into the cornea1 stroma (arrows). In the case of the Elwax implant alone or containing BSA, no detectable capillary proliferation takes place. In the case of the implant containing either FGF or EGF, capillaries proliferate from the limhus t,oward the implant.
506
D. GOSPODAROWICZ,
H. BIALECKI
ASD
T. K. THAKRAL
With 1 pg of EGF, a positive response was olbserved in 40:;) of the cases, while with 2.5 and 5 pg it was observed in 60 and 70”; of the cases respectively. In all positive cases. capillaries started to invade the cornea1 stroma 3 -4 days following iml~lant~ation and reached the implant within 8 10 (lays. invading 2-3 mni of cornea1 stroma. In all cases, t,he capillaries had the grc)ss morphological appearance of an elongated hairpin progressing toward the implant. No capillaries were ohservccl within the corneai stroma in regions far away from the implant [Fig. 2(a)]. In 110 case was inflammation detectable, as reflected by the lack of cornea1 edema or the absence of an hypopyon in the anterior chamber. The cornea stayed clear at all times. With concentrations of EGF ranging from 10 to 60 pg per implant, a positive response was observed in lOOy, of the cases, Capillaries invading the cornea1 stroma started to appear as early as two days after implantation and developed into a brush border which reached the implant within five clays. They afterwards developed into 2 or 3 main Lvascular trunks which regressed 4-G weeks later, probably because of the depletion of EGF in the implant. The histological appearance of corneas implantecl
FIG. 3. Histological examination 14 days post implantation of the efi’ect of an Elwax 40 implant containing 10 CL::of EGF on the structure of the cornea. (a) ~lomoal pocket. The epithelium over thts cornea1 pocket shows clear signs of hyperplasia ( :: 40). (b) ( ‘entral part of the cornea. No cpithelial hyperplnsia is ohserved (x40). (c) Same as (a) but showing the cornea1 epithelium (X 100). (d) &NW as (b) but showing the cornea1 epithelium ( :< 100). (a) D i ff erentiated blood vessels ( ~200). (f) Blood vessels infiltrating the cornea1 stroma (x 100).
FIBROBLSST
AND
EPIDERMAL
GROWTH
PAC’TOR
50s
Fro. 4. Presence of polymorphonuclear eosinophilic leukocytes in the stroma of cornea carrying an Elwas 40 implant containing either 5 pg of EGF (a) ( ‘I\ 100) or 40 pg of BYA (b) ( x 100). In both cases. nrovascularization of the cornea1 stroma developed after a period of 3 days. While with the EGF implant few or no inflammatory cells could be seen, (only eosinophylic leukocytes, could be detected (a) arrows, : 400). In the case of an Elwax 40 implant containing BSA, there was a massive infiltration of t,h+t stroma by polymorphonuclear leukocytes (b), ( .< 100).
508
D. GOSPODAROWICZ,
H. BIALECKI
AND
T. K. THAKRAL
by slow-release form of EGF is shown in Fig. 3. The cornea1 epithelium located over the cornea1 pocket containing the implant, was distinctly thicker than that located far away from the implant [Fig. 3(a), (b)]. While the cornea1 epithelium consisted of a basal cell layer and three layers of wing cells [Fig. 3(d)] in the part of the cornea remote from the implant, in the region overlying the implant as many as 16 layers of wing cells could be seen on top of the basal cell layer [Fig. 3(c)]. The morphological appearance of capillaries growing into the stroma is shown in Fig. 3(e), (f). The absence of inflammatory cells such as mononuclear cells and pseudoeosinophilic leukocytes, which are conspicuous in the rabbit because of their tinctorial properties, is evident. Only in rare cases were inflammatory cells present, and when present [Fig. 4(a)] only a few (less than 5) could be detected in each section (Table I, II: III). In some cases: inflammation could be detected, as was reflected by the presence of macrophages and other inflammatory cells: but tjhe total number of these cells was never higher than lo6 per whole cornea, as was determined hy counting inflammatory cells in cornea1 sections [Fig. 4(b)]. These corneas were recorded as inflamed (Table I) and were not scored as positive for either FGF or EGF. TABLE
Number of cornea tested
10 12
The angiogenic
PC’F T concentration
400 ‘40
II.
Neovasculerization Positive Kegstive
5 4
3 6
Inflammation Positive Ntyative
7 1
3 6
e#ect of slozu release form of FGF
When slow-release form of FGF ranging in concentrations from 0.5 to 6Op.g was implanted in the cornea1 stroma, a positive response was observed with concentrations ranging from 2.5 to 6Opg (Table II). No response was observed at a concentration of 0+5pg. The response observed with an implant containing 1 pg of FGF was in one case attributed to a local inflammation (Table II). With 2.5p.g of FGF a positive response was observed in SO:/, of the cases, while
FIBROBLAST
AKD
EPIDERMBL
GROWTH
FACTOR
509
with 5 pg of FGF it was observed in 85% of the cases. With either concentration, the neovascularization can be considered to be a weak response, since a response was not seen before day 4 and reached the implant only 4-8 days later. As already observed with low concentrations of EGF, the capillaries had the gross morphological appearance of an elongated hairpin. In one of 44 cases, a pronounced inflammation was observed. Histological examination of the corneas revealed the presence of macrophages and inflammatory cells in only 3 of 44 cases. These corneas were scored as presenting signs of inflammation (Table II). Wit,h concentrations of FGF ranging from lo-60 pg in the implantation, a posit’ive response was achieved in 100% of the cases, and, as with high EGF concentrations, two days follow-ing the implantation the corneas exhibited capillaries growing into the cornea1 stroma as a brush border and reaching the implant 4 days later [Fig. 2(b)]. ,2s wit’li the EGF implant, in rabbits which were not sacrificed, the capillaries began to regress 4-6 weeks later.
FIG. 5. Histological examination, 14 days post implantation, of the effect of an Elwax 40 implant containing 10 pg of FGF on the structure of the cornea. (a) Cornea1 pocket, partly jelled with red blood cells. (b) Region of the cornea adjacent to the limbus. Blood vessels can be seen growing in the upper part of the cornea1 stroma (arrows) (x20). (c) Same as (a), but showing the cornea1 pocket at a higher magnification (;i 100). (d) Same as (b), but showing the upper part of the cornea1 stroma where capillaries are present at a higher magnification ( x 100). (e) Capillaries and cornea1 stroma ( x 200). (f) A group of capillaries and blood vessels ( x 400).
510
D. GOSPODAROWICZ,
H. BIALECKI
ASD
T. K. THAKRAL
The histological studies of corneas with implants containing FGF are shown in Fig. 5. In contrast to corneas implanted with EGF in Elwsx 40, there \vas no epithelial hyperplasia of the cornea1 epithelium underlying the cornea1 stroma. nor was there any significant number of eosinophil leukocytes or mononuclear cells. In some cases, the cornea1 pocket which had no time to completely seal co~~ltl he seen to fill witjh red blood cells when vessels reached it.
When the angiogenie effect of I&ax 40 alone s-as tested in the cornea, it was found to be inactive. In 10 cases out of 10. it, did not induce any vascularization of t,he cornea (Table III). Histological examination of the corneas show that. the Elwax 40 implant~s did not induce any epithelial hvperplasia nor did they seem to have an! significant, effect on the kemtocytes of the cornea1 stroma [Fig. 6(a). (b), (c)].
Fx. 6. IIistologicsl ennminetion, 14 days post implantstion, of the effect of an E1w1.u 40 implsut on the structure of the cornea. (a) Corned pocket ( \: 20). (b) Region of the cornea betwcen the corncal pocket and the limbus. No blood vessels arc prcsrnt ( x 20). (c) Corned pocket seen at, u higher magnification (x 100).
When Elwax implants containing BSA were implanted in the stroma, either 40 or 400 pg of protein failed to induce vascularization in six cases out of 10. and three cases out of 10 respectively (Table III). In the four cases (40 yg BSA) and seven cases (400 pg B&4) where vascularization was observed, there was a massive infiltration of leukocytes (mononuclear cells) into the cornea1 stroma. Numerous inflammatory cells (eosinophylic leukocytes) and macrophages could be observed. 4. Discussion Capillary proliferation has been shown to be a general feature of actively growing tissue, whether they are of normal origin, such as the corpus luteum (Gospodarowicz and Thakral, 1978), salivary gland (Hoffman, McAushan, Robertson and Burnett, 1976), and granulation tissue (Algire and Chalkley, 1965), or from tumors, such as
FLHRORLAST
APJD EPIDEKMSL
GROWTH
FACTOR
511
solid malignant tumors represented by the Brown Pearce epithelioma, the V, carcinoma, B16 mouse melanoma, and by the sarcoma 37, (Folkman, 1974; Folkman and Cotran, 1976). Neovascularization can also be induced by extracts from various cell types derived from either tumors (Phillips, Steward and Kumar, 1976), established cell lines such as Balb/c 3T3: (Klagsburn, Knight,on and Folkman, 1976). or from neutrophils (Fromer and Klentworth, 1976). It has been demonstrated that’ at. least in the case of tumor tissue one can purify an angiogenic factor (TSF) which. when implanted into the cornea in the form of a slow-release form of a pol,vmcr consisting of an acrylamide pellet, induces neovascularization in vivo (Folkman. !Merler, Abernathy and Williams. 1974; Phillips, St’eward and Kumar, 1976). No studies on the angiogenic potency in vivo of mitogens such as the epidcrrnal or fibroblast growth factors have been conducted. although there is a dist’inct possiIjility t(hat these factors could be angiogenic in viva. For example, FGF has ljeen shown to l)e a strong mitogen for cultured vascular endothelial cells. EGF is present, at high concentrations in mouse submaxillary glands and crude extracts from this gland are angiogenic in vivo (Folkman and C’otran. 1976). In addition, Savage and ( ‘ohen reported in 1972 that the topical administration of EGF to corneas denutletl of their epit’helium resulted not only in a rapid reepithelization of the denuded area but also induced the blood vessels in the limhal region to become much more promin ent in t’he EGF-treated eye than in the control. It is conceivable? therefore, that although EGF does not have a pronounced mitogenic eflect on vascular endothelial cells in vitro. it’ could have an effect, on neovascularization in viva, either directSly OI indirectly. When the anpiogenic properties of both EGF and FGF are testecl in viva, using for clelivcry a slowrelease form polymer composed of Elmax 40 implanted in the rabtnt cornea (a tissue normally transparent and avascular), our results indicate that’ t)oth mitogenic agents induce capillary proliferation. The optimal concentrat,ion of mitoprns required for such an effect is in the range of 10 pg. When tested at sul)optimxl concentrations, EGF is a more potent angiogenic agent than FGF. Although the concentrations of mitogens used in these studies are 100. to lOOO-fokl higher than bhose required in vitro t,o see a mitogenic effect on nlesotlerm-derivetl cells, one has to keep in mind that it is unlikely, when one uses a slow-release forni p(Jlpler such as Elwax 40, that all the mitogen will be delivered at once at the limbal region. ;Iccording to the data of Langer and Folkman (1976), after the mibogcnic agent is released from the polymer in an initial burst, it is later released at minutc~ concentrations but at a sustained rate. and this process can last for weeks. The concentrahion of mitogen which will diffuse through the cornea1 stroma on a pr diew basis will in all likelihood be much smaller than the original amount, put into the polynicr. One of the pit~falls in the study of neovascularization in the cornea has been the dilhcultv of distinguishing between the straight angiogenic effect of a mitopenic agent and the induced inflammatorv process caused by tissue extracts or organic agents (polymer) inserted into the cornea1 stroma. This inflammatory process has been recognized in all cases to be the most potent inducer of angiogenesis and therefore represents a potentially troublesome source of artefact in studies of cornea1 capillary proliferation. Although it has been considered by some that the clarity of the cornea, the lack of cornea1 edema. and the lack of hypopyon in the anterior chamber could reflect a lack of inflammation. these are only signs which may also occur in t’he acute phase of the inflammation process. It has been shown by others that, although a cornea can stay perfectly clear, an invasion by polymorphonuclear
512
D. QOSPODAROWIC‘Z,
H. BIALECKI
AND
T. K. THAKRAL
cells can precede a phase of vascularization (Fromer and Klintu-orth, 1976). It has also been shown that activated macrophages implanted into the cornea1 stroma are a source of factor(s) Rhich can induce vascularization (Clark, Stoney, Leung, silver, Hohn and Hunt, 1976; Polverine, Cotran, Gimbrone and Unanue, 1977). For these reasolls, therefore, a study of tbe angiogenic potential of mitogens implanted in the cornea1 stroma should be accompanied by a histological study of the cell types present during the active phase of capillary proliferat.ion. Although in this study the presence of inflammatory cells was not examined during the initial phase of capillary proliferation (first. seven days) but only 7-14 days after implantation. our results demonstrate that. in the case of corneas containing either an FGF or EGF implant, an invasion by polymorphonuclear leukocytes and nlacrcJphages is unlikely to be the cause of the long-term capillary proliferation. In all cases. these cell types were either ahsent from the cornea1 stroma, particularly around the cornea1 pocket where they should accumulate, since it is the densest region of capillaries, or, when present, their total concentration was far t)clow that used tjv others to induce neovascularizat,ion. In the case of inflammation, for esample, a massive invasion of t,he whole cornea1 stroma by polymorphonuc1ea.r leukocytes has been reported by Framer and Klintworth (1976)> while in the case of luaorophages, no less than 5 x 1Oj cells are required for an angiogenic response (Clark et al.. 1976; Polverine et a,l., 1977). That t’he angiogenic properties of FGF or EGF are distinct from those of leukocyte- or macrophage-mediated events can be further deducecl from the observation that, when BSA is llsed, in every case where one ohserves vascularization, it correlates with a massive invasion of the cornea1 strorna 1)~ leukocytes. No such correlat,ion could be observed when t,he neovascularization was induced hv e&her FGF or EGF. This observation tends to demonst.rate t.hat two diflerent types of neovascularization can be observed in vivo, one of which is induced hp a specific and purified mitogen acting at low concentrations. This type of vascularization is not mediated by leukocytes or rnacrophage,,res. In contrast. the neovascularizat,ion induced by non-specific agent)s, such as BSA. which has no nlitogcnic a&iv+per se but is capable of inducing an infiammation. could \JC niediatctl 1)~ leukocyt~cs anal macrophages. Finally, one has to resolve whether the neovAscLllarizati(jII inclucetl in viva I)y either FGF or EGF ha,s a direct or indirect effect on the vascular endot~helium. In the case of FGF: which has been shown to be mitogenic for vascular entlothelial cells maintained in vitro, the possibility exists that. in vivo it could directly stimulate the proliferation of capillary endothelial cells. With EGF‘, such a direct effect is 1)~ no mea,ns evident. First, EGF has been shown to be either inactive or poorly rnitogenic for bovine va,scular endothelial cells and human endothelial cells, respectively. Therefore, there is no reason why it should act directly on the capillary endothelium in vivo. Secondly, EGF has been shown to be a potent mitogen for the cornea1 epithelium, which responds to it with a striking hyperplasia. It is therefore possible that the neovascuIarization observed in the case of EGF is not a direct effect of the mitogen but is secondary to the cornea1 epithelial hyperplasia. Alternatively, EGF could directly trigger the proliferation of capillary endothelial cells: although it had little effect on this cell type in vitro.
FIBROBLAST
AND
EPIDERMAL
GROWTH
FACTOR
513
ACKNOWLEDGMENTS
This work was supported by grants from the NH (EY 01068 and NICHD-11082) to Dr Gospodarowicz and (GM-12828) to Dr T. Hunt. W e wish to thank Mr Harvey Scodel for his invaluable assistance in the preparat,ion of this manuscript.
REFERENCES Algire, G. H. and Chalkley, H. W. (1945). Vascular reactions of normal and malignant tissues in vivo I. Vascular reactions of mice to wounds and t,o normal and neoplastic transplants. Nufl. Canc,er Inst. 6, 75-85. Clark, R. A., Stoney, R. D.. Leung, D. Y. K., Silver, I., Hohn, D. C. and Hunt, T. K. (1976). Role of macrophages in wound healing. Surg. Forum 27, 16-19. Folkman, J. (1974). Tumor angiogenic factor. Cnncer Res. 34, 2109-23. Folkman, J. and Cotran, R. (1976). Relation of capillary proliferation to tumor growth, In RP~‘. Exptl. Pathol. (Ed. Richter, G. W.). Pp. 207-48. Fromer, C. H. and Klintworth, G. I
