basic fibroblast growth factor stimulates 3h-thymidine...

Investigative Ophthalmology & Visual Science, Vol. 31, No. 7, July 1990Copyright © Association for Research in Vision and Ophthalmology

Basic Fibroblasf Growth Factor Stimulates °H-ThymidineUptake in Retinal Venular and Capillary

Endothelial Cells in VivoEugene de Juan,* Einar Srefansson,t and Akihiro Ohira*

Recent studies have found basic fibroblast growth factor (bFGF), an angiogenic peptide, in retina andhave suggested that bFGF is responsible for retinal vascular proliferation. To test the hypothesis thatbFGF stimulates 3H-thymidine uptake in retinal vascular cells in vivo, we injected bFGF (100 ng) intothe vitreous cavity of six cats at 0 hr and again at 24 hr. Eight control eyes received boiled bFGF or noinjection. After 46 hr, 3H-thymidine was injected into the vitreous cavity of all eyes and 2 hr later theeyes were enucleated. Intense 3H-thymidine uptake was seen in eyes with bFGF (56 ± 20 SD positivecells per section) but not in control eyes (7-10 positive cells per section (P < 0.001). Trypsin digestpreparations showed that the thymidine uptake was predominantly in the venular (89%) and capillary(10%) endothelium and not in arterioles (1%) (P < 0.001). The data suggest that retinal venularendothelial cells respond preferentially to exogenous bFGF, and in part may explain their prominentrole in the neovascular process. In a second group of experiments to test the hypothesis that retinalischemia releases a diffusable factor similar to bFGF that can cause 3H-thymidine uptake in retinalvascular cells, we created branch retinal vein occlusion in six cat eyes. The fellow eyes received noinjections. In the eyes with branch vein occlusion there was an intense 3H-thymidine uptake within thedistribution of the occluded vein (84 ± 77 SD positive cells per section), but none in the areas outsidethe occluded vein (P < 0.001). The data suggested that branch vein occlusion does result in a signifi-cant expression of endothelial cell mitogenic activity within the distribution of the occluded vein, butthis activity did not appear to be mediated by a diffusable factor that increased mitogenic activity in theretina outside the occluded vein's distribution. Invest Ophthalmol Vis Sci 31:1238-1244,1990

Retinal neovascularization in proliferative diabeticretinopathy, retinal vein occlusion, sickle cell ane-mia, and retinopathy of prematurity is hypothesizedto be mediated by a release of angiogenic factors fromischemic retina.1"3 This hypothesis is attractive be-cause it appears to explain new vessel growth outsidethe distribution of a branch vein occlusion or theincreased risk of iris neovascularization after removalof the lens and vitreous in patients with diabetic reti-nopathy.4"6 The finding that normal retina containssubstantial amounts of an angiogenic peptide, basicfibroblast growth factor (bFGF), supports this hy-pothesis.78

bFGF is an 146-amino-acid polypeptide that ispresent in retina and is widely distributed in the

From Duke University Eye Center,* Durham, North Carolinaand the University of Iceland.f

Supported by the National Institutes of Health grants EY-05903and EY-07001 and by Research to Prevent Blindness, Inc.

Submitted for publication: February 15, 1989; accepted No-vember 27, 1989.

Reprint requests: Dr. Eugene de Juan, Box 3802, Duke EyeCenter, Durham, NC 27710.

body.9 In several bioassays it is reported to be angio-genic even at low (1-10 ng/ml) concentrations.9"" Itcauses endothelial cell mitosis, chemotaxis, and inva-sion into collagen gels in vitro.12"15 According to onereport, elevated amounts of bFGF are present in vit-reous of diabetic patients.16

If bFGF serves such a function in retihovasculardisease, it should be able to elicit a proliferative re-sponse from the retinal vasculature. To test this hy-pothesis we injected bFGF in the vitreous cavity ofcats and measured 3H-thymidine uptake by the reti-nal vasculature. In a related experiment, we tested thehypothesis that retinal ischemia releases an angio-genic growth factor similar to bFGF, by creatingbranch retinal vein occlusions in the cat and studyingthe proliferative response both within the distributionof the occlusion and outside the area of the vein oc-clusion.

Materials and Methods

Cats (2-4 kg) of both sexes were used in the experi-ments and were treated according to ARVO Resolu-tion on Animal Experimentation. Under intramus-

1938

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933157/ on 06/04/2018

No. 7 bFGF STIMULATES RETINAL VENULES / de Juan er al 1239

cular ketamine (20-40 mg/kg) and xylazine (2mg/kg) anesthesia, 100 ng recombinant bovine bFGF(Amgen, Mountain View, CA) in 0.1 ml phosphatebuffered saline (PBS) was injected through the parsplana (4 mm behind the limbus) temporally using atuberculin syringe with a 27-gauge needle into theright eye of six cats. The left eye was used as an un-treated control. Three eyes were injected with 100 ngbFGF that had been boiled for 3 min and dissolved in0.1 PBS as a control. After 24 hr the procedure wasrepeated.

In a second group of six cats, branch retinal veinocclusions were produced in the right eye by placing a20-gauge diathermy probe (OMS, Danvers, MA)through a pars plana sclerotomy and by using an indi-rect ophthalmoscope to place the probe over the reti-nal veins (superior and nasal only).17 Then by apply-ing current (intensity level 2) in the diathermy unit(OMS), the vein was occluded usually with one ap-plication. The arteries were avoided.

After 46 hr, 10 nC 3H-thymidine (specific activity20 Ci/mM; New England Nuclear, Boston, MA) wasinjected into the vitreous cavity of all eyes. Two hourslater the cats were sacrificed by an overdose of intra-cardiac phenobarbital and the eyes removed. After a24-hr fixation in buffered formalin, the anterior seg-ments were removed at the pars plana and a 10-mmhorizontal section containing the optic nerve wasprepared for paraplast embedding and autoradiogra-phy. The retina from the superior collotes was re-moved and trypsin digests were prepared.18 For au-toradiology, 5-/im paraplast retinal sections were cutor the trypsin digests were placed on gelatin coatedslides. The slides were then immersed in Kodak(Rochester, NY) NTB-2 photographic emulsion andallowed to dry under red-light conditions. The slideswere placed in a light-tight black slide box with desic-cant and stored in the dark for 7 days and then devel-oped, fixed, and stained.

3H-thymidine uptake of unstained retinal crosssections was counted by two observers who were un-aware of the surgical manipulation of the eyes; thesecounts then were averaged. The results are expressedin terms of counts per section with a standard devia-tion calculated.

3H-thymidine uptake was quantified in the trypsindigests by counting all available tissue and expressingthe counts as positive cells per low-power field. Ineyes with a branch vein occlusion, counts were madeseparately in areas corresponding to the distributionof the occluded vein(s). These counts were made by asingle observer who was aware of the surgical manip-ulations, because the distribution of the occludedvein needed to be known for accurate comparisons.

Similarly, the ratio of positive cells in venules, cap-illaries, and arterioles was calculated after counts ofmore than 100 positive cells in each section had beenreached. Results are expressed as cell number and astandard deviation.

Results

Effects of bFGF Injected Intravitreallyinto Normal Cat Eyes

Clinical observations: Injection of 100 ng bFGFinto the normal cat vitreous appeared to cause noimmediate effect. There was no immediate dilation ofthe vasculature or evidence of inflammation. After 24hr there again was no noticeable change in the eye.The injection was repeated. Forty-six hr after theoriginal injection of bFGF there appeared some milddilation of the retinal vasculature. Six bFGF injectedeyes did not appear clinically different than the threeeyes injected with boiled bFGF.

Histologic and autoradiographic findings: The tis-sue sections of the posterior segments of the normalcat eyes fixed 48 hr after the injection of bFGFshowed no apparent histologic abnormalities. Therewas no retinal edema or evidence of inflammatoryinfiltration into the retina in any of the eyes injected.

Autoradiographic studies of the retinal tissue sec-tions from the six uninjected eyes showed only mini-mal uptake of 3H-thymidine in the retina or retinavasculature (7 ± 5 positive cells per 5-^m retinal sec-tion, mean ± SD). This was also true of the threeboiled-bFGF-injected eyes. In these eyes we averaged10 ± 1 positive cells per section. However, in the sixeyes that received active bFGF there was a 5-foldincrease in the number of labeled cells in the tissuesections (average 56 ± 20 positive cells per section). Astudent t-test comparing the boiled bFGF and activebFGF injected eyes revealed that this result washighly statistically significant (P < 0.001). In the tis-sue sections, many labeled cells, which were locatedin the nuclear layers, appeared to be vascular endo-thelial cells.

Trypsin digest preparations ofbFGF-injected eyes:In an effort to isolate the surrounding retinal tissuefrom the retinal vasculature, trypsin digests were pre-pared in three normal eyes, in the three boiled-bFGF-injected eyes, and in three active-bFGF-in-jected eyes. Autoradiographic preparations of thetrypsin digests allowed a straightforward determina-tion of capillaries, venules, and arterioles. Labeledcells were distinguished easily. Trypsin digests of theeye that had not received bFGF or had receivedboiled bFGF had very little label in the vascular cells(Fig. 1). However, in the eyes that received active


1240 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / July 1990 Vol. 31

'„ , - • * " ' - AFig. 1. Autoradiographtc prepara-

tion of trypsin digest from cat retina 48hr after intravitreous injection ofboiled bFGF. No uptake of 3H-thymi-dine is evident (X100):

$m • 1

4

4t

Fig. 2. (A) Autoradiographic preparation of trypsin digest fromcat retina 48 hr after intravitreous injection of recombinant bovinebFGF (100 ng). Heavy uptake of 3H-thymidine is evident in ven-ules and to a lesser degree in the capillaries (X40). (B) Higher-powermagnification of a venule in the inner retina {XI60).

bFGF there was an intense labeling of the vascula-ture, primarily in endothelial cells (Figs. 2A, B). Todetermine the distribution of the labeled cells, we re-corded the location of over 100 cells in each of thepreparations. The location was subdivided into arte-rioles, capillaries, and venules. The results were aver-aged and tabulated. We found that 89% of the labeledcells were located in the venules and to a much lesserextent in the capillaries. The arterioles took up almostno label (Fig. 3).

Branch Vein Occlusion Studies

Clinical observations; The six eyes in which branchvein occlusions were induced developed immediatevenous engorgement and by 48 hr showed intrareti-nal hemorrhage and retinal edema and occasionallyserious retinal detachment.

Histologic and autoradiographic observations: Inthe eyes with branch retinal vein occlusion, therewere varying degrees of inner retinal edema and cellloss consistent with inner retinal ischemia.

Autoradiography of tissue sections of the involvedsegments of these eyes showed a heavy labeling of theinner retina (Fig. 4). Many of the sections containedlabeled vessels that were surrounded by a diffuseinner retinal edema (Figs. 5A, B).

To quantitate the labeling we counted the labelledcells in the areas where there was histologic evidenceof branch retinal vein occlusion (ie, inner retinaledema) and expressed the counts as number of la-beled cells per XI00 field. We found that within the

i


No. 7 bFGF STIMULATES RETINAL VENULES / de Juan er ol 1241

ARTERIOLES CAPILLARIES VENULES

Fig. 3. Relative distribution of vascular cells with 3H-thymidineuptake in trypsin digest in eyes with bFGF injected into the vitre-ous cavity. The venules contain 89% of label (P < 0.001).

distribution of the occluded vein there was a veryhigh labeling index (83 ± 77 SD labeled cells perXI00 fluid field; n = 6).

Trypsin digest preparations of eyes with branchvein occlusions: The trypsin digest preparations ineyes with branch vein occlusion were not as free ofretinal tissue as the bFGF-injected eyes. We specu-lated that this was due to trypsin inhibitors in serumthat had leaked from the vessels. However, the digestswere sufficient to determine the distribution of thelabel in the vasculature. We found in the areas ofbranch vein occlusion that venules and capillaries

contained approximately equal amounts of label.However, as we found with the bFGF-injected eyes,the arterioles contained less than 1 % of the label(Fig. 6).

To determine the effect of branch retinal vein oc-clusion of the uptake of 3H-thymidine in areas out-side or distant to the distribution of the occludedvein, we counted labeled cells in the distribution ofthe occluded vein (n = 6) and outside of these areas inthree eyes. We found that within the distribution ofthe occluded vein there was a high labeling index ofthe venules and capillaries (48 ± 30 cells per low-power field; n = 6). In marked contrast, we foundoutside of the distribution of the occluded vein thatthere was a very low labeling index (1 ± 1 cell perlower power field; n = 3). This difference was highlystatistically significant {P < 0.001). When comparedto the findings of the bFGF-injected eyes, we foundwithin the distribution of the occluded vein that thelabeling index approached that of active bFGF. How-ever, outside the occluded vein the labeling index wasnormal (Fig. 7).

Discussion

The marked increase in 3H-thymidine uptake inthe venular and capillary endothelial cells in the pres-ence of bFGF suggests that bFGF may play a role inretinovascular proliferation. The data shows a veryintense labeling in the venular endothelium afterbFGF injection. Within 48 hr, large portions of thevenules had areas in which the endothelial cells wereheavily labeled. In the trypsin digest preparations, the

Fig. 4. Autoradiogram showing mod-erate 3H-thymidine uptake in the innernuclear layer of retina after branch veinocclusion. Inner retina shows mild cysticedema (no bFGF injection) (X100).


1242 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / July 1990 Vol. 31

Fig. 5. (A) Autoradiogratn of crosssection of retina after branch vein occlu-sion (no exogenous bFGF). Note 3H-thymidine-labeled cells in distendedvein (X100). (B) Higher-power magnifi-cation of labeled vessel in inner retina(X400).

B

ARTERIOLES CAPILLARIES VENULES

Fig. 6. Relative distribution of vascular cells with 3H-thymidineuptake in trypsin digest in eyes with experimentally induced branchvein occlusion (no bFGF injection). The capillaries and venulescontain nearly all label.

veins contained 89% of all of the label. One possibleconclusion from this data is that the veins respondedpreferentially to bFGF. This correlates well with theclinical and histologic observations that neo vascularbudding usually begins from the retinal venules.3'19

In the eyes with a branch vein occlusion only, therewas a moderate uptake of 3H-thymidine. The oc-cluded vein caused a "blood-and-thunder" appear-ance and sometimes was associated with an exudativeretinal detachment. We have previously demon-strated that the preretinal pO2 is quite low over ex-perimentally induced branch retinal vein occlusions(7 torr, whereas normal = 21 ton).20 Therefore, wereasoned that if ischemic retina does release a solubleangiogenic growth factor into the vitreous cavity,then one might expect that experimentally inducedbranch vein occlusion in the cat would release such acompound. We further reasoned that if the com-pound had the same or similar properties as bFGF,


No. 7 bFGF STIMULATES RETINAL VENULES / de Juan er al 1243

80

z

TR

YP

S

z

-J

LJJ

oUJ

LISO

Q.

Q

UJ

WE

R F

OQ.

O

DCUJai -

IGE

<

Q

7 0 -

6 0 -

50 -

40 -

3 0 -

2 0 -

10 -

\s / / / /\

I

NORMAL BOILED FGF BRVO BRVO

FGF IN OUT

Fig. 7. Number of labeled cells per lower power field in trypsindigest preparation of eyes that were normal, that received boiledbFGF (100 ng) intravitreally, or that received active bovine bFGF(100 ng). The last two columns show the results of counting thenumber of cells in eyes with a branch retinal vein occlusion(BRVO) either within the distribution of the occluded vein (IN) oroutside the distribution of the occluded vein (OUT).

bFGF plays, if any, in normal retina not undergoingmitogenic activity.

Several explanations have been proposed. One isthat endogenous bFGF is sequestered in the extracel-lular matrix, presumably to bound to heparan sulfatein the basement membrane. Moscatelli has demon-strated that there are two major binding sites forbFGF;21 one is a high-affinity receptor that is cell-as-sociated and the other is a low-affinity matrix store.The matrix stores are accessible to the cell as it needsit. Why the large stores of bFGF do not normallycause a mitogenic response remains unexplained.Another possibility is that bFGF in a tissue is nor-mally14 in an inactive form that in some way becomesactive when needed. A third possibility is that theregulation of cell proliferation lies not with the con-centration of a mitogenic growth factor, but ratherwith the cell itself. It is more than conceivable that thecell regulates its receptors and then responds to thesurrounding sea of growth factors when needed.

Key words: basic fibroblast growth factor, capillary endo-thelial cells, retina, venules, 3H-thymidine

then we should see a similar response in the normalretinal vasculature as was caused by exogenousbFGF. However, we found that the branch vein oc-clusion did not cause any uptake of 3H-thymidineoutside the distribution of the vein occlusion. Thisfinding was consistent with a series of experiments werecently completed demonstrating that there is littleor no increase in extractable endothelial cell mito-genic activity between branch vein occluded andnormal retina.20

Endothelial cell uptake of thymidine was seen pre-dominantly in venules and to a lesser degree in capil-laries. Arteriolar endothelial cells did not take upthymidine in response to bFGF injection. This is astriking difference between the arteriolar vascularcells and the venular and capillary endothelial cells.The observation is consistent with those of Michael-son3 and Williams et al,19 showing that blood vesselgrowth in the retina originates from the venous end ofthe microcirculation (ie, the postcapillary venules).Our data suggest that the venous side of the microcir-culation is particularly sensitive to the exogenousgrowth factor.

Although this study demonstrates that bFGF caninduce thymidine uptake and presumably cell prolif-eration in vascular and nonvascular cells in the ret-ina, this does not mean that fibroblast growth factoris necessarily the regulatory factor in vascular prolif-eration. It remains to be demonstrated what role

Acknowledgments

The authors are grateful to Trilby McClammy for excel-lent secretarial assistance.

References

1. Patz A: Clinical and experimental studies on retinal neovascu-larization. XXXIX Edward Jackson Memorial Lecture. Am JOphthalmol 94:715, 1982.

2. Wise GN: Retinal neovascularization. Trans Am OphthalmolSoc 54:17, 1956.

3. Michaelson IC: The mode of development of the vascular sys-tem with some observations on its significance for certain reti-nal diseases. Trans Ophthalmol Soc UK 68:137, 1948.

4. Finkelstein D, Clarkson J, and Diddie K: Branch vein occlu-sion: Retinal neovascularization outside the involved segment.Ophthalmology 89:1357,1982.

5. Aaberg TM: Clinical results in vitrectomy for diabetic tractionretinal detachment. Am J Ophthalmol 88:246, 1979.

6. Blankenship G, Cortez R, and Machemer R: The lens and parsplana vitrectomy for diabetic retinopathy complications. ArchOphthalmol 97:1263, 1979.

7. Glaser BM, D'Amore PA, Michels RG, Patz A, and Fenslau A:Demonstration of vasoproliferative activity from mammalianretina. J Cell Bio 84:298, 1981.

8. D'Amore PA, Glaser BM, Branson SK, and Fenselau AH:Angiogenic activity from bovine retina: Partial purificationand characterization. Proc Nat Acad Sci USA 78:3068, 1982.

9. Folkman J and Klagsburn M: Angiogenic factors. Science235:442, 1987.

10. Esch F, Baird A, Ling N, Ueno N, Hill F, Denoroy L, KlepperR, Gospodarowicz D, Bohlen P, and Guillemin R: Primarystructure of bovine pituitary basic fibroblast growth factor(FGF) and comparison with the amino-terminal sequence of


1244 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / July 1990 Vol. 31

bovine brain acidic FGF. Proc Nat Acad Sci USA 82:6507,1985.

11. Lobb RR, Alderman EM, and Fett JW: Introduction of angio-genesis by bovine brain derived class 1 heparin-binding growthfactors. Biochemistry 24:4969, 1985.

12. Thomas K, Candelore M, Gimenez-Gallego G, DiSalvo J,Bennett C, Rodkey J, and Fitzpatrick S: Pure brain-derivedacidic fibroblast growth factor is a potent angiogenic vascularendothelial cell mitogen with sequence homology to interleu-kin. Proc Nat Acad Sci USA 82:6409, 1985.

13. Gospodarowicz D, Cheng J, Lui G, Baird A, Esch F, and Boh-len P: Corpus luteum angiogenic factor is related to fibroblastgrowth factor. Endocrinology 117:2383, 1985.

14. Gospodarowicz D, Massoglia S, and Cheng J: Isolation of pitu-itary fibroblast growth factor by fast protein liquid chromatog-raphy (FPLC): Partial chemical and biological characteriza-tion. J Cell Physiol 12:323, 1985.

15. Montesano R, Vassali JD, Baird A, Guillemin R, and Orci L:

Basic fibroblast growth factor induces angiogenesis in vitro.Proc Nat Acad Sci USA 83:7297, 1986.

16. Baird A, Culler F, Jone K, and Auillemin R: Angiogenic factorin human ocula fluid. Letter. Lancent 2:562, 1985.

17. Chan CC, Rice T, and Green W: Experimental occlusion of theretinal vein. Graefes Arch Clin Exp Ophthalmol 224:507,1988.

18. Cogan D, Toussaint D, and Kuwabara T: Retinal vascularpatterns. Arch Ophthalmol 107:1589, 1968.

19. Williams J, de Juan E, and Machemer R: Ultrastructuralcharacteristics of new vessels in proliferative diabetic retinopa-thy. Am J Ophthalmol 106:986, 1988.

20. de Juan E, Stefansson E, and Dickson JS: Capillary endothelialcell mitogenic activity in experimental branch vein occlusion.Graefes Arch Clin Exp Ophthalmol 228:191-194, 1990.

21. Moscatelli D: Metabolism of receptor-bound and matrix-bound basic fibroblast growth factor by bovine capillary endo-thelial cells. J Cell Bio 107:753, 1988.


basic fibroblast growth factor stimulates 3h-thymidine...

Documents