Vascular Development in Primate Retina:Comparison of Laminar Plexus Formationin Monkey and Human
Ray F. Gariano,* M. Luisa Iruela-Arispe,"\ and Anita E. Hendrickson*-f
Purpose. The temporal and spatial sequence of development of laminar vascular plexuses wasdetermined qualitatively and quantitatively in monkey and human retina.
Methods. Histologic and cytochemical methods were used to study Macaca monkey eyes fromfetal day 55 (F55d; birth = F168d) to 17 years, and human retina from fetal 21 weeks toadult.
Results. In monkey retina, spindle-shaped, presumed vascular precursor, cells appear at F55din the nerve fiber layer (NFL) adjacent to the optic nerve. The vascular plexuses in the NFL-ganglion cell layer appear first and form in the presence of spindle cells. Nerve fiber layervessels extended radially to reach the temporal ora at F95d and nasal ora at FllOd. Thecapillary plexus at the inner border of the inner nuclear layer (INL) appears at F120d nearthe optic disc, whereas the plexus at the outer INL border appears at F130d. Both reach theirfinal position before birth. The INL plexuses form by endothelial budding from more vitreadvessels in the absence of spindle cells. In the NFL, vessel growth to match retinal growth atthe ora also involves endothelial budding. The growth rate of all plexuses was approximately225 //m/day. The central fovea and the most peripheral retina adjacent to the ora serrataremained avascular throughout development. Differences between humans and monkeysinclude: Human vessels complete maturation after birth; human vessels reach the nasal oraearlier than the temporal ora; and spindle cells are more abundant and dispersed over agreater area within human NFL. Growth rates of human plexuses were comparable to thosein monkeys.
Conclusion. In both primates, deeper capillary plexuses form only by extension from existingvessels (angiogenesis). In the NFL, early vessel formation involves spindle precursor cells(vasculogenesis). The main difference between monkey and human in these processes is thatthe mature monkey vascular pattern is established well before birth. Invest Ophthalmol VisSci. 1994; 35:3442-3455.
JL he intrinsic vasculature of adult human and mon-key retina consists of inner capillary beds within thenerve fiber layer (NFL) and ganglion cell layer (GCL)and two outer capillary beds along the inner and outerborders of the inner nuclear layer (INL). The lattertwo vascular networks have been respectively named
From the Departments of* Ophthalmology and f Biological Structure, University ofWashington, Seattle, Washington.Supported by National Institutes of Health grants EY07031 and EY04536, in partby NIH grants EY01730 and RR00166, and by an award from Research to PreventBlindness, Inc. AEH is a Senior Scholar of Research to Prevent Blindness and aResearch Affiliate of the RPRC. MLJ-A is supported by American Heart Associationgrant 93011970.Submitted for publication December 7, 1993; revised March 1, 1994; acceptedMarch 4, 1994.Proprietary interest category: N.Reprint requests: Ray Gariano, MD, PhD, Department of Ophthalmology, RJ-10,University of Washington, Seattle WA 98101-0001.
the shallow (SINL) and deep (DINL) plexuses.1 TheNFL and GCL vessels also include arterioles and ve-nules, whereas the other laminae are comprised exclu-sively of capillaries. Each laminar plexus runs in anearly planar direction, and these are interconnectedby vertical capillary segments.2 The distribution ofthese vascular laminae varies along the central-pe-ripheral direction of the retina. Most of the centralretina is supplied by vessels in all four vascular lami-nae. The SINL and DINL plexuses disappear in themidperiphery, and only a single layer of NFL-GCLvessels is present in the far periphery. Vessels are ab-sent altogether in the fovea centralis, and the far pe-ripheral retinal rim adjacent to the ora serrata. Thesefeatures of mature primate retinal vasculature aredemonstrated in Figure 1.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
Blood Vessel Development in Monkey Retina 3443
B
FIGURE l. Vertical sections through retina of a P2w monkey showing vascular plexuses. Moreperipheral retina is to the right in all photomicrographs. (A) Nasal peripapillary retinacontains NFL, GCL, SINL, and DINL vascular layers (labeled as 1, 2, 3, and 4, respectively).Arterioles and venules are located within NFL and GCL only, whereas capillaries are foundin all laminae. (B) Midperipheral retina. Note termination of DINL vessel plexus (arrowhead).The NFL and GCL plexuses are often difficult to distinguish in peripheral retina and arelabeled as a single plexus (1/2). (C) In far peripheral retina, only vitread vessels remain.This plexus terminates (arrowhead), leaving avascular retina adjacent to ora serrata. (D)Vessels are absent within the foveal depression (right of the arrowhead). Glycol methacrylatesection stained widi azure II and methylene blue. Calibration bar =100 fxm.
The development of these vascular plexuses is ofinterest for several reasons. First, the pattern of intra-retinal vessels recapitulates the horizontal arrange-ment of the retina into neuronal, plexiform, andnerve-fiber layers and thus, corresponds to the orga-nization of retinal functions. Second, a vision-threat-ening disease in newborn infants, retinopathy of pre-maturity, is associated with retinal vascular develop-ment and may occur in part because of the specialorganization of retinal vessels.3 Pathologic changes inthis condition may be restricted to certain vascularlaminae.4'5 Third, prior studies of monkey6'7 and hu-man4'8"10 retinal vascular development contain littlequantitative data regarding vessel growth and often donot evaluate the vascular plexuses as separate entitiesduring development. Finally, previous reports are con-tradictory on the mechanism of foveal vascular devel-opment.6'7
The present study was undertaken to determineboth quantitatively and qualitatively the temporal andspatial sequence of developmental events that culmi-
nate in adult primate retinal angioarchitecture. Em-phasis was placed on the description of individual vas-cular plexuses. The Old World Macaca monkey retinawas studied because retinal development in this spe-cies closely parallels that in humans, and it has beenwidely considered to be an appropriate model for hu-man retina development11 We also studied a series ofhuman fetal, neonatal, and infant retinas12 to obtaina direct comparison of the temporal and morphologiccharacteristics of retinal vascular formation in bothprimates.
SUBJECTS AND METHODS
Monkey eyes were from Macaca nemestrina or M. mu-latta monkeys ranging in age from fetal 55 days (F55d;normal gestation 165 to 170 days) to adulthood. Fulldentition is present by 5 years, and the oldest eyestudied was 17.7 years. M. nemestrina retinas were stud-ied at F55d, F60d, F7ld, F75d, F83d, F90d, F95d,F106d, F108d, F119d, F132d, F136d, F147d, F157d,
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
F165d; postnatal day 1 (Pld); postnatal week 2 (P2w),P4w, P6w, P16w; postnatal month 9 (P9m); and post-natal year 10 (PlOy) and P17.7. M. mulatto, eyes wereavailable at F97d, F125d, F140d} and F155d. All tissuewas obtained in accordance with the guidelines estab-lished by the Regional Primate Research Center at theUniversity of Washington and with the ARVO State-ment for the Use of Animals in Ophthalmic and VisionResearch. Prenatal retinas were obtained from fetusesthat were the result of timed pregnancies (M. nemes-trina; ± Id) or were staged by experienced personnelusing a comparison of weight, crown-rump, femur,and foot length against a colony standard (M. mulatto.;± 5d).
Fetuses were delivered by aseptic cesarean sectionand given an overdose of barbiturate. Postnatal ani-mals were sedated with ketamine followed by deepanesthesia with barbiturate. The eyes were then enu-cleated, opened, and postfixed for 1 to 3 hours. In oneseries of M. nemestrina animals, one eye was enucleatedimmediately after the barbiturate took effect, the cor-nea and lens were removed, and the posterior globewas immersed in 4% paraformaldehyde fixative for 1to 3 hours. The entire horizontal meridian of theseeyes was cryoprotected and serially frozen-sectioned at10 fim on a cryostat. The other eye also was enucleatedimmediately and injected with methyl carnoys (60%methanol, 20% chloroform, 10% acetic acid) fixativewithout opening and immersion-fixed overnight. Inthese eyes, the horizontal meridian was paraffin em-bedded and serially sectioned at 6 //m. In anotherseries of M. nemestrina, the animals were perfused with4% paraformaldehyde containing 0.1% to 0.5% glu-taraldehyde, the eyes were enucleated and dehy-drated, and the entire horizontal meridian was thenembedded in glycol methacrylate. Blocks were seriallysectioned at 2 fim, and every third section wasmounted for staining.
Human tissue was obtained through the Lions'Eye Bank at the University of Washington and wasavailable for ages F21w, F25w, F27w, F28w, F34w, Pld,Plw, P22m, and P13y. Eyes were enucleated within 3hours of death, and the posterior half was immediatelyimmersed in 4% paraformaldehyde/0.5% glutaralde-hyde at least overnight. The horizontal meridian wasdehydrated, embedded in glycol methacrylate, and se-rially sectioned at 2 /xm, and every third section wasmounted for staining. Only parafoveal horizontal sec-tions were used in this study.
Histology
Paraffin and frozen sections were stained using cresylviolet, Masson's trichrome, and hematoxylin and eo-sin. Plastic sections were stained with methylene blue-azure II. Endogenous peroxidase activity within eryth-rocytes was used to identify vessels with blood flow. Todetect endogenous peroxidase, frozen sections were
N F L , v •>,. " o 'i» - » "' *•!• +7**j£dk
HGURE 2. (A) Spindle-shaped cells in the NFL adjacent tothe optic nerve head at F60d (arrow). (B) Peripheral to theoptic nerve, clusters of spindle cells (arrow) lie just abovecystic spaces (*) within the NFL (F71d). (C) Spindle cellsform cord-like structures (arrowheads) continuous with pa-tent vessels. Note DAB-positive red blood cells in the lumen(arroio) and lack of cystic spaces in NFL. Glycol methacrylatesection stained with azure II and methylene blue. Calibra-tion bars = 70 /xm (A), 18 ^m (B), 35 /xm (C).
washed with phosphate-buffered saline (PBS) for 5minutes, and paraffin-embedded sections were depar-afinized before washing in buffer. All sections werethen incubated in 0.03% diaminobenzidine in 0.1 MTris buffer for 30 minutes to label endogenous redblood cell catalase activity, washed in cold tap waterfor 10 minutes and, in some sections, counterstainedwith cresyl violet.
WholemountsRetinal wholemounts from eyes at F90d, F105d toFllOd, F120d, F140d, P3w, and P9y were processed
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
Blood Vessel Development in Monkey Retina 3445
for ADPase cytochemistry.13 Briefly, retinas wereteased from the underlying pigment epithelium andchoroid shortly after enucleation. Radial cuts were cre-ated peripherally to allow flattening of the retina, andthe tissue was fixed for 3 hours in 4% paraformalde-hyde at 4°C. Retinas were washed in 0.1 M PBS andthen in cold tap water to remove fixative, and theywere incubated for 20 minutes at 37°C in a reactionmedium containing 0.2 M Tris maleate buffer, (pH7.2), 6 raM MgCl2, 3 mM PbNO3, and ADP 1 mg/ml.Fetal eyes were incubated for 1 hour because imma-ture vessels stain lighter.13 The tissue was washed fivetimes with distilled water and incubated in 2% ammo-nium sulfide for 1 minute at room temperature beforewashing with distilled water. Tissue was then floatedonto a glass slide and coverslipped with glycerin.
RESULTS
Monkey Retinal Vessel Development
No consistent difference was noted between retinas ofdifferent Macaca species, so the results will be reportedas a group. Our observations were limited to sectionsthrough the optic nerve head parallel to the hori-zontal meridian. The superior and inferior far periph-ery were not available for analysis. Sections weremapped for the presence and location of blood ves-sels, for retinal structural features that might be re-lated to angiogenesis, and for presumed vascular pre-cursor cells identified by their location within theNFL-GCL, their spindle shape, and their alignmentinto cords and/or tube-like structures near the ad-vancing edge of angiogenesis.1415 The distribution ofNFL, SINL, and DINL vascular plexuses across a ma-ture monkey retina is shown in Figure 1.
At F55d to F65d, spindle-shaped cells in the NFLwere present adjacent to the optic nerve, but theyextended no farther than 500 //m peripherally in bothnasal and temporal directions (Fig. 2a). The origin ofspindle cells could not be determined in our material.No budding of spindle cells from the hyaloid vesselsor from any foci within the nerve was observed, norwere precursor cells such as angioblasts identified. Nopatient vessels were seen at F55d to F60d in the retina.Spindle-shaped cells in the NFL were seen only inoccasional sections, suggesting that these cells extendin a discontinuous pattern not apparent in transversesections.
Intraretinal vessels first appear in the NFL atabout F70d when they extend 500 to 700 /xm temporalto the optic nerve. Characteristically, spindle cellswere found 300 to 750 jum peripheral to the advancingvessels, suggesting that they could be formative ele-ments for developing vessels. Evidence for this in-cludes the observations that spindle cells extendedfiber-like processes into the NFL parallel to their long
FIGURE 3. The first vessels (anow) apparent at F71d in themonkey are in temporal NFL near the optic nerve, just cen-tral to a cluster of spindle cells. (A) Formation of deeperplexuses proceeds by an extension of vessels from the NFL-GCL to the SINL (arrow) at F125d. Glycol methacrylate sec-tion stained with azure II and methylene blue. Calibrationbar = 30 /xm.
axis (Fig. 2b) and that occasionally they were alignedin cord-like structures contiguous with vessels with apatent lumen (Fig. 2c). It is possible that such struc-tures are not solid cords but are vessels whose lumensare out of the plane of sectioning. Although spindlecells were most numerous just peripheral to the ad-vancing wave of newly forming vessels, they were occa-sionally seen in the central NFL far from the site ofangiogenesis. Spindle-shaped cells were even foundinfrequently in adult retina near the optic nerve, butclusters of cells or aligned adjacent cells were onlyfound during the period that NFL and GCL vesselswere growing most rapidly (F70d to F120d),
By F75d, vessels extended from the nerve approxi-mately 1500 /xm temporally and 1200 fim nasally (Fig.3a). Once vessels appeared in the NFL, they rapidlyprogressed into peripheral retina so that they reachedwithin 200 to 600 ^m of the ora serrata by FlOOdtemporally and FllOd nasally. Even though the retinacontinues to grow throughout late fetal and early post-natal life, the vessels remained a constant distance
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
FIGURE 4. Graphic summary of linear growth of monkey vascular plexuses along the parafo-veal horizontal meridian as a function of fetal and postnatal age. Growth is expressed aspercent of the distance from the optic nerve head to the ora serrata for the most peripheralvessel of each plexus. Gray = NFL vessels; white = SINL vessels; black = DINL vessels.
from the ora so that the peripheral 200 to 600 //mnear the ora serrata is never vascularized.
At F95d, vessels appear in the GCL temporal tothe optic disc, and by FllOd they are seen in the GCLnasal to the disc. Vessels in the GCL rapidly extendacross the retina and reach their adult position justcentral to the farthest peripheral extent of the NFLvessels 7 to 14 days after their appearance. Becausethe NFL and GCL vessels are virtually coextensive andbecause it is often difficult to distinguish betweenthem in peripheral retina where the GCL is thin anduneven (Fig. 1), these two lamina will be consideredas a single "vitread" plexus of vessels.
Vessels in the SINL were first seen at F119d intemporal and F125d in nasal retina (Fig. 3b), but it isunclear whether they arise first in the retina adjacentto the optic nerve or around the fovea. The last vascu-lar plexus to form was the DINL, which appeared atF132d near the optic disc on its temporal side. As withthe vitread vessels, SINL and DINL vascularization be-gan centrally and extended radially, but it did notreach as close to the ora serrata and terminated in-stead in the midperiphery. Vertically oriented capil-lary segments were often seen extending from the NFLand GCL vessels to the SINL or DINL (Fig. 3b), aswell as from the SINL to the DINL. Spindle-shapedcells that were found in the vitread retina were neverassociated with capillary growth in the SINL or DINL.
The temporal sequence of linear growth of NFL-GCL, SINL, and DINL vascular plexuses is summa-rized by the graph in Figure 4. Note that: (1) Vasculari-zation in monkey retina proceeds in a stepwise fashion
with each plexus developing in the sequence NFL >SINL > DINL; (2) vitread (NFL-GCL) vessels reachthe ora by about 60% of gestation (F105d); (3) theadult pattern of laminar vasculature is reached shortlybefore birth (F155d to F165d); and (4) vessels appearfirst on the temporal side of the disc, vascularizationproceeds slightly faster in the temporal versus nasaldirection, and NFL-GCL vessels reach the temporaledge of the retina first.
The distances from the optic nerve head to thenasal and temporal ora serrata and to the most periph-eral extent of each vascular lamina were measured inhistologic sections to determine the rate of retinal andvascular growth. During the period F75d to F150d,the length of the horizontal meridian increased in anearly linear fashion at a rate of approximately 48/im/day on the temporal side and 58 //m/day on thenasal side (Table 1). Vessel growth occurred in aroughly linear fashion at early ages but quicklydropped to a rate approximately equal to that of reti-nal growth (Fig. 5). The rates of growth during thefast linear phase were calculated for each vascularplexus, and the results are given in Table 1. Vesselsgrew at a rate roughly fourfold greater than the rateof overall retinal growth. The average growth rate forall plexuses was 224 //m/day (range, 202 to 268 fim/day). In general, vascular extension was more rapidon the temporal than on the nasal side and was slightlyfaster for the vitread than for the inner nuclear layervascular plexuses. The most rapid growth occurred inthe temporal NFL vascular plexus.
Cystic spaces have been described in the puppy
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
Blood Vessel Development in Monkey Retina 3447
100 150POST-CONCEPTION AGE IN DAYS
200
FIGURE 5. Plot of the linear extent of each vascular plexus(in mm) as a function of age, taken from monkey nasalretina. Each plexus initially grows in a rapid linear phasethat quickly levels off. Lines are derived by regression analy-sis. Open boxes = NFL; diamonds = SINL; closed boxes =DINL.
NFL in the region of newly developing vessels andspindle-shaped precursor cells,14"16 which are thoughtto provide a structural framework in which angiogen-esis occurs.15 We observed similar spaces in monkeyNFL. These were more abundant and well-defined atF71d to FllOd when NFL vessels were radiating fromthe optic nerve to the ora serrata, but they were notpresent in every retina. In monkey retina, cystic spaceswere typically found peripheral to the most distal spin-dle cells (well in front of the advancing wave of angio-genesis), were frequently absent in areas of presumedvessel growth (Figs. 2c, 3a), and were occasionally seenin regions of relatively mature vessels. When cysticspaces and spindle-shaped cells presumed to be pre-cursor cells were found in the same region, the cellsdid not occupy a consistent position within the spacesbut were instead adjacent to the internal limiting
membrane (Fig. 2b) or were in the deeper NFL orGCL. An alternative explanation is that these are fixa-tion artifacts in this fetal tissue.
Vascular development in the entire retina was ex-amined in wholemounts at F90d, F105d to FllOd,F120d, F140d, P3w, and P9y, according to a methodfor ADPase cytochemistry reported by Lutty andMcLeod13 and previously used to assay ATPase activityby Flower et al.14 The area of the wholemounted retinawas drawn using a camera lucida, and the peripheralextent of each vascular plexus was determined at mul-tiple circumferential points (Fig. 6). In this article, wereport results obtained from these wholemounts thathelp explain the sequence of vascular laminar devel-opment. Vasculogenic and angiogenic mechanismsshown in the wholemounts will be the subject of asubsequent article.
In general, wholemount data confirmed our ob-servations in retinal sections but contributed new de-tails of topographic development. Figure 6 demon-strates that blood vessel development in monkeys isrelatively symmetric around the optic disc. Blood ves-sels exist only in the NFL and GCL of central retinaup to F105d to FllOd. SINL vessels appear around theoptic disc by F120d and approach their final distribu-tion in the periphery by F140d. The DINL vessels atF140d are still growing, although even in the adultthe DINL vessels only reach the midperiphery. Notethe difference in retinal size, especially between theF140d and the adult retina. This indicates that theretina enlarges after birth, which requires the bloodvessel plexuses also to expand in a commensuratemanner.
Retinal wholemount ADPase cytochemistry pro-vided a more detailed view of the formation of thedeeper plexuses. At F120d in central retina, short cellu-lar extensions or buds arise from die vitread vessels andextend into the SINL (Fig. 7a). Usually two buds arise
Growth rates of monkey and human retina and NFL, SINL, and DINL vascular plexuses, in microns per day. The given age range refersto the period during which growth occurred in a linear fashion. Data for humans are available only for the temporal retina. CC =Correlation coefficient, describing fit of growth over time to linearity (P < 0.05 for all coefficients given); F = fetal; P = postnatal.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
from adjacent vitread capillaries; these adjacent exten-sions lengthen and then connect to form arches (Fig.7b). Neighboring arches then interconnect, again bycellular extensions, to form the planar SINL capillaryplexus. At F140d, this budding process was occurringmore peripherally in the SINL and had begun in theDINL of central retina. The arches varied in length,but their height in a given region was similar, assuringthat the deeper capillary plexuses would have a planarconfiguration. Interconnections between vitread anddeeper plexuses developed almost exclusively throughthe formation of arches in that single, long processesrunning between plexuses were rarely seen.
The area destined to become the fovea in theadult can be identified by F55d on the basis of conemorphology, retinal lamination, and GCL thickness.17
In horizontal sections, the foveal NFL and GCL weredevoid of spindle cells and vessels from the earlieststages of development onward. Wholemounts demon-strated a more complex pattern of foveal vascular de-velopment (Fig. 6). In the F90d wholemount, it canbe seen that the foveal avascular zone begins as thenasal rounded end of an avascular band that extendstemporally along the horizontal meridian to the oraserrata. This avascular band becomes vascularized byextension from superior and inferior sides of the ra-phe. Vessels never invade the nasal-most region, leav-ing an avascular circular region centered on the fovea.Vessels also are slow to fill in the temporal edge ofthe raphe, often leaving an avascular notch later indevelopment. The raphe is vascularized by F120d.
At the margin of the fetal foveal avascular zone,short bud-like extensions from the vessels point intoward the center of the fovea. These resembleaborted attempts at vessel extension and were presentup to and including F140d, the oldest of the fetalwholemounts in this study. In postnatal retina, vesselsbordering the foveal avascular zone lack these budsand, instead, end in smooth loops that form a perifo-veal capillary ring. A nearly identical situation is seenin retina adjacent to ora serrata. After NFL-GCL ves-sels have reached their final position within 200 to600 fj,m of the ora serrata around FlOOd to FllOd, theretina continues to grow at least to 2 years of age.18
During retinal growth, the vessels at the ora extendbud-like processes toward the avascular zone. In ma-ture retina, after growth has ceased, retina adjacentto the ora contains smooth vascular loops that anasto-mose to form the peripheral vascular arcade. A com-parison of vascular development at the margins of theavascular fovea and ora serrata is shown in Figure 8.
Choroidal Vascular Development
Patent blood vessels containing red blood cells wereevident in the monkey choroid from the edge of theoptic nerve to the ora serrata well before intrinsicretinal vessels appeared. At F51d, a single-layer of
choriocapillaris was present adjacent to the retinal pig-ment epithelium across the entire retina, but thestroma deep to the choriocapillaris contained fewlarger vessels (Fig. 9a). At F60d to F75d, the choriocap-illaris is more densely packed with vessels, and me-dium and large-size stromal vessels are increased innumber (Figs. 9b, 9c). By F88d, the choroid contained"numerous larger vessels in the outer stromal portionand a more tightly packed choriocapillaris across thefull retinal extent (Fig. 9c). There was a tendency fordeep choroidal development to be more advancednearer the optic nerve head, but this was much lessmarked than for intraretinal vascular development. Inlate prenatal and early postnatal ages, the choroidthickens because of an increased density of capillariesin the choriocapillaris and a larger number of mediumand large vessels in the inner and outer stroma (Statt-ler's and Haller's layers, respectively). Pigmentation isfirst apparent in the outer stroma within a few daysafter birth, and then increases progressively within thestroma until adulthood (Figs. 9d to 9g).
Human Retinal Blood Vessel Development
Vascularization in human retina showed marked simi-larities to that described above for the monkey. Theearliest process was a central-to-peripheral radial de-velopment of the vitread vascular plexuses thatseemed to involve spindle-shaped precursor cells. AtF21w, the youngest age examined, vessels had ex-tended 2 to 3 mm from the optic nerve head bothnasally and temporally. Spindle-shaped cells were scat-tered in the NFL from the optic nerve to approxi-mately 500 //m peripheral to the distal-most vessels.The most peripheral spindle cells were often foundin clusters and occasionally were aligned to form cord-like structures adjacent to or contiguous with newlyformed vitread capillaries. This was followed by theappearance and radial extension of SINL at F28w andthen of the DINL at F30w. Both plexuses developedin the absence of any apparent precursor cells, similarto monkey SINL and DINL plexus development.Growth of each of the human retinal vascular plexusesis summarized in Figure 10.
Analysis of human temporal retinal dimensionsduring the period F21w to Plw revealed a linear rateof retinal growth somewhat greater than that in themonkey (82 //m/day versus 48 fim/day). During thesame period, the temporal NFL vascular plexus exhib-ited a biphasic growth curve, with a growth rate of 281/zm/day from F21w to F28w and 94 //m/day from F28wto Plw. The temporal SINL and DINL plexuses alsogrew in a linear fashion during their periods of great-est growth at rates of 177 //m/day and 193 //m/day,respectively. These values are comparable to those cal-culated for the monkey (Table 1).
Despite the overall similarities between humanand monkey retinal vascular development, several dif-
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
Blood Vessel Development in Monkey Retina
F90d F105d F120d
F140d P9y
FIGURE 6. Schematic depiction of the peripheral-most extension of each vascular plexus atvarious ages, based on ADPase stained wholemounts. White = Avascular; black = NFL vesselsonly; diagonal lines = NFL + SINL vessels; gray = NFL + SINL + DINL vessels; arrowheadindicates optic nerve; temporal is to the right in all drawings. Note the avascular zone onthe horizontal meridian at F90d, which extends from fovea to ora. This area is vascularizedexcept for the fovea and a small, far peripheral wedge by F105d.
ferences were noted. First, in human retina at theearliest stages of formation, NFL vessels extended fur-ther into nasal than temporal retina. For instance, atF21w, NFL vessels extended 1.89 mm temporal and3.44 mm nasal to the optic nerve head. This observa-tion is consistent with several reports that vessels reachthe nasal ora serrata approximately 1 month earlier
than they do on the temporal side.19 Second, spindle-shaped, presumed precursor cells in the NFL weregenerally more numerous and more easily found inhumans. Third, human retinal vasculature was lesswell developed at birth than monkey vasculature.Monkey retina had a complete NFL vascular plexusby FlOOd to FllOd (60% to 65% of gestation), and
FIGURE 7. ADPase-stained monkey rednal wholemounts. (A) Cellular processes extendingfrom the NFL (out of focus) to the level of the SINL (in focus). (B) Extensions from NFLto SINL form arches. Calibration bar = 50 fim.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
FIGURE 8. ADPase-stained monkey retinal wholemounts. (A) At F120d, cellular processesextend toward the avascular peripheral rim (ora serrata at top). (B) Ora serrata of adultexhibits continuous anastomosing perioral arcade. (C) Foveal avascular zone (F) at F120d;note cellular extensions from perifoveal vessels toward the avascular zone (arrow). (D) Fovealavascular zone in the adult exhibits a continuous perifoveal capillary ring. Calibration bar= 100 j/m.
all plexuses attained the adult configuration by aboutF155d (92% gestation). In contrast, the NFL plexusin the human does not even reach the ora until nearterm in the temporal retina, and the major portionof the SINL and DINL plexuses develop between F34w(85% gestation) and Plw (Figs. 4, 10).
The human choroid appeared mature at F21w,with a well-formed choriocapillaris and medium andlarger vessels in the inner and outer stromal layers.Pigmentation appeared in the outer layer around birthand dien increased progressively for several years.
DISCUSSION
In most organs, vascular development is achieved bytwo basic mechanisms.20 The first entails de novo for-mation of new blood vessels from vascular precursorcells and is termed vasculogenesis. The second pro-cess, angiogenesis, involves growth of new vessels bybudding or extension from preexisting vessels. Previ-ous reports in feline,21 canine,15 rat,22 and human8"10
retina suggest that mammalian retinal vascularizationuses both developmental mechanisms. This report
comparing monkey and human retinal vessel develop-ment supports this conclusion. In primates, the twomechanisms are spatially and temporally separate. Thefirst vascular plexuses to form are located in NFL andGCL and appear to arise, at least in part, from matura-tion and fusion of spindle-shaped precursor cells, orvasculogenesis. The later-developing deeper plexusessurrounding the inner nuclear layer develop in theabsence of these precursor cells, and they grow byextension from previously formed vitread vessels usingclassic angiogenic mechanisms.
Although we cannot exclude the possibility thatspindle-shaped cells seen in our embryonic retinal sec-tions subserve some other function, we suggest thatthey are vascular precursor cells because they are typi-cally found in advance of developing vessels, they aremost numerous during the period of vasculogenesis,they appear to form cord-like structures that align andconnect with patent vessels, and they are similar inmorphology to vascular precursor cells identified inpuppy, kitten, rat, and human retina.4'14"16"21'22
Whether these cells originate by differentiation of insitu angioblasts15 or arise from optic nerve head vessels
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
Blood Vessel Development in Monkey Retina 3451
"s*lPSSr_ •-/•' ̂ _ "
' •#1
FIGURE 9. Choroidal vascular development in the monkey. (A) F51d, continuous choriocapil-laris {arrowheads) and occasional larger stromal vessels are present. (B) F60d, increasednumber of medium-sized midstromal vessels (arrow) are evident. (C) FlOOd, choriocapillaris,medium-sized vessels, and outer stromal large vessels are present. (D) F151d, no pigmentis visible. (E) F165d, small amount of pigmentation present within stromal melanocytes{arrow). (F) Plm, denser stromal pigmentation. (G) Adult (PlOy), further increase in pig-mentation. Glycol methacrylate section stained with azure II and methylene blue. Calibrationbars = 50 (im (A, B), 57 /im (C, D), 26 ^m (E, F).
and migrate into the retina16 could not be determinedin these histologic sections.
Because astrocyte migration from the optic nerveinto the retina also coincides with vasculogenesis,23 itis possible that the spindle-cell population is heteroge-neous and includes both astrocyte and vascular pre-cursors. In cat retina, spindle cells can be distin-guished from glial cells by immunohistochemicalmarkers,21 but in primates cell markers have not yetbeen identified that separate spindle cells from other
cell types in the NFL. In preliminary experiments, wehave been unable to label monkey spindle cells withantibodies to vascular markers, including von Wille-brands factor, laminin, and Ricinus lectin, but havefound occasional monkey spindle cells labeled by anti-bodies to the glial marker glial fibrillary acidic pro-
tein.24
Monkey retina vessels grow up to 268 ^m/dayduring development, yet spindle cells are seen only inoccasional sections, and most newly developing vessels
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
FIGURE 10. Graphic summary of linear growth of human vas-cular plexuses in temporal retina as a function of fetal andpostnatal age. Growth is expressed as percent of the distancefrom the temporal border of the optic nerve head to theora serrata for the most peripheral vessel of each plexus.Black = NFL vessels; diagonal lines = SINL vessels; gray =DINL vessels.
at the advancing front of vascularization were not obvi-ously associated with them. It seems unlikely that sofew precursor cells could form the entire NFL-GCLvascular plexus in the absence of intense proliferation,yet mitotic figures were not observed in our material.If some spindle cells are immature astrocytes, thiswould reduce the vascular precursor pool even fur-ther. Our data imply that other mechanisms besidesspindle cell-mediated vasculogenesis are involved invessel formation in the primate NFL-GCL and thatthe role of spindle cells may not be direcdy related tovessel formation.
Intraretinal vessels were first noted adjacent to theoptic nerve head of the monkey at E70d. Vitread ves-sels extended to the ora serrata within approximately30 days temporally and 40 days nasally so that by FlOOdto FllOd, the entire retina is supplied by a monolayerof vitread vessels. The first tier of deeper capillariesforms from about F120d in the peripapillary SINL,whereas the final tier at the DINL is first seen some10 days later. Retinal vascular maturation generallyfollows a radial vector that also describes many otherdevelopmental sequences, such as onset and cessationof cytogenesis, neuronal differentiation, glial migra-tion and maturation, and synaptic formation.11'25"27
This same sequence of blood vessel development,from central to peripheral and from vitread to sclerad,has been described in the cat,21 dog,14 mouse,28 rat,16
and human.810
During the periods of maximal growth, vessels ineach lamina extend at a rate approximately threefoldto fivefold greater than the rate of retinal growth;this results relatively quickly in an adult-like retinaldistribution of each plexus. In each lamina, vesselsextend slighdy faster on the temporal side as pre-
viously suggested.8'9 It is interesting that SINL andDINL vessels grow at a similar rate, which was slighdyslower than the rate for vitread vessel extension. Thismight reflect distinct rates of spread by vasculogenicversus angiogenic mechanisms. Human vascular plex-uses exhibited comparable linear growth rates to thosefound in monkey. These growth rates refer only toone-dimensional anterior spread of the plexus borderand ignore differences in the total capillary volumebetween plexuses and volume changes within the plex-uses secondary to remodeling during development.
The primate retina differs from that of cat, dog, andrat in that it has a highly developed fovea with an avascu-lar center. Previous studies in monkeys have reportedconflicting accounts of the development of the macularcirculation. Henkind et al,7 using an ink injectionmethod, found vessels within the central fovea early indevelopment that later regressed to form the foveal avas-cular zone. Another study using PAS staining of retinalwholemounts6 found an avascular foveal center from theearliest stages of development. Kirby29 likewise observeda vascular free zone in the fovea throughout develop-ment when retinal wholemounts were examined withNomarski optics. Our wholemount labeling with ADPasestaining supports the concept that the foveal center isnever vascularized.
In addition, we found that the foveal avascularzone does not begin as a discrete isolated avasculararea. Instead, in early gestation, a horizontal avascularband extends along the horizontal meridian from thefovea to the temporal ora serrata. This band appearsto be a watershed between advancing inferior and su-perior vascular arcades. By F105d, the temporal por-tion of the band gradually fills with vessels, leaving acircular avascular zone centered on the fovea. Theavascular horizontal band probably does not resultfrom failure of branches of the superotemporal andinferotemporal arcade vessels to meet in the hori-zontal meridian because no such avascular band existsnasally, despite the presence of superonasal and in-feronasal arcades. Rather, some active mechanismmust exclude vessels from the fovea. This avascularpattern is an exception to the rule that vascularization,like other retinal developmental processes,27 proceedsin a continuous central-to-peripheral direction. Weare unaware of any distinctions in neuronal differenti-ation along the temporal and nasal horizontal meridiathat could account for this difference.
A second avascular zone exists at the most periph-eral margin of retina along the ora serrata, which alsoremains free of vessels during development. As theretina itself grows throughout much early life, the ves-sels along the margins of the ora serrata exhibit strand-like extensions into the avascular area, which we thinkis vascular growth in response to increasing retinalarea. Because spindle cells were not noted near theora serrata after FlOOd to FllOd in vertical sections,
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
Blood Vessel Development in Monkey Retina 3453
vessel growth near the ora after this age appears todevelop by a process of angiogenesis. Thus, both angi-ogenesis and vasculogenesis contribute at differenttimes to form portions of the vitread vasculature. Asimilar situation may exist adjacent to the foveal avas-cular zone, where spindle cells are absent and wherebud-like processes extend centripetally from vessels atthe foveal margin. In infancy, vessels encircling thefovea and those adjacent to the ora anastomose toform a continuous capillary border.
Retinal vascular development has been extensivelystudied in the cat by Chan-Ling and Stone.21 These inves-tigators found that vitread vessels in the cat also developin association with precursor spindle cells and thatdeeper plexuses and perifoveal capillaries subsequentlydevelop from vitread vessels by a budding process. Thecat also exhibits in development an avascular band tem-poral to the central macula (area centralis). Unlike thatin primates, however, the area centralis ultimately be-comes vascularized in cats. The cat also differs fromprimates in that the deeper plexuses extend peripherallyfrom the area centralis; in primates, all plexuses initiallyarise adjacent to the optic nerve head.
The stimulus for development of intrinsic retinalvessels remains obscure. It has been suggested thatincreasing retinal thickness during development low-ers tissue oxygen concentration in the superficial ret-ina by increasing the diffusion distance to the choroidand thereby stimulates vascularization.3'30"32 However,factors other than retinal thickness must be involvedin primate retinal angiogenesis because the correla-tion between retinal thickness and vascular supply isimperfect.1 For example, during development thecontiguous avascular band extending temporally fromthe fovea is actually thicker than adjacent vascularizedretina, yet this area remains devoid of blood vesselsfor some time.
Snodderly and colleagues' related the distributionof enzymes involved in oxidative metabolism (cyto-chrome oxidase33'34 and malic dehydrogenase35) inadult monkey retina to the topographic distributionof capillary plexuses. This implies that neuronal andglial metabolic activity, especially in the inner retinallayers, better accounts for local tissue oxygen require-ments and levels than retinal thickness alone. In fact,in several species studied, those with high oxidativemetabolism in the inner retinal layers contained anintrinsic vasculature, whereas those with low oxidativemetabolism did not.34'35 From a developmental stand-point, inner layer metabolism in turn reflects theemerging physiological capacity of developing retinalcells. Interestingly, synaptic vesicle protein 2, a proteininvolved in neurotransmitter release from synaptic ves-icles that may indicate potential for synaptic activity, isexpressed in a central-to-peripheral gradient roughlycoincident with the development of vitread retinal ves-sels.26 Thus, tissue oxygen tension depends on the
interplay of retinal thickness, anatomic compartmen-talization of metabolic cellular machinery, and topo-graphic onset of electrical and other retinal activities.
An interesting possibility is that tissue hypoxia me-diates vessel growth by elaboration of angiogenic fac-tors. Basic fibroblast growth factor and vascular endo-thelial growth factor (VEGF) are present in the ret-ina.28'36"38 Vascular endothelial growth factor mayguide the extension of developing brain vessels fromthe pial to the ventricular surface,39 a process thatis similar embryologically to angiogenic extension ofvitread retinal vessels into deeper lamina. Expressionof VEGF is increased under hypoxic conditions in situand in cultured endothelial cells,40'41 thus makingVEGF an attractive candidate for a mediator of vesselgrowth during retinal development.
Factors other than tissue metabolism and oxygenlevel could also contribute to retinal vascularization.Initial vasculogenic growth of vitread vessels is associ-ated with spindle-cell appearance and maturation, fol-lowed by endothelial cord and lumen formation.These processes may depend on factors such as thebiochemical composition of the vitread retina that fa-cilitates spindle-cell differentiation42'43 and the priordevelopment of glial structures that mechanically sup-port or otherwise direct vessel extension.15'44 Prelimi-nary studies in developing rodent retina found thatvascular precursor cells migrate along a fibronectinmeshwork that may be produced by astrocytes.4445 Inother tissues, fibronectin has been implicated as aguide for vessel extension.20
Elucidation of the factors involved in develop-mental retinal vascularization may also clarify abnor-mal vascularization in disease states. Some prematureinfants exhibit aberrant retinal vascular growth, lead-ing to retinopathy of prematurity.19 Diseases in whichretinal ischemia is prominent, such as diabetic reti-nopathy, are characterized by an undesirable prolifer-ation of vessels on the vitread retinal surface.46 Prelimi-nary studies have implicated basic fibroblast growthfactor, VEGF, and other factors in the pathogenesisof abnormal vessels in proliferative retinopathy,3847
and animal models of retinopathy of prematurity re-vealed altered glial structure and glial-vascular rela-tionships.48 An important direction in the study ofprimate vascular development will be to determinespecific factors that promote vascularization and torelate these factors to the pathogenesis of retinal neo-vascular diseases. Conversely, future research mightrelate the cellular and molecular developmental sig-nals that regionally inhibit or temporally terminatevascularization processes to potential therapies thatsuppress pathologic neovascularization.46'49
Comparison With Human RetinalVascularizationThe basic plan of retinal vascularization is similar inmonkeys and humans, with sequential central-to-pe-
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
ripheral growth of vitread followed by deeper retinalplexuses. Formation of the foveal avascular zone alsoappears similar in both species in that this area isdevoid of spindle cells and vessels at all ages examined.However, the timing of vascular plexus growth differsbetween species. Vitread vessels reach the ora serrataby about 65% gestation in the monkey, whereas hu-man temporal peripheral retina is vascularized justbefore term.89 Deeper plexuses have reached theirperipheral-most extent by about 92% gestation in themonkey but do not reach the ora until the first postna-tal week in humans. Earlier vascular development inmonkeys may relate to their earlier maturation of reti-nal neuronal elements.11 The completion of retinalvascularization before birth makes the monkey retinaa poor model for human neonatal retinal vascularpathologies. The major portion of retinal vasculariza-tion in lower mammals occurs after birth.1416'2128
Spindle-shaped cells appeared in advance of de-veloping vitread vessels and exhibited alignment andcord formation in both monkey and human retina.Both human and monkey vascular plexuses grew in anearly linear manner during their periods of greatestgrowth, and the rates of growth of vitread and of SINLand DINL plexuses for the two species during theseperiods were comparable. Michaelson10 reported a"very approximate" growth rate of 100 /zm/day forsuperficial retinal vessels in the human fetus, but thisvalue was averaged over periods of varying growthrates. It is unclear why that study found a triphasicgrowth curve when our specimens exhibited a mono-phasic linear growth rate during the major portion ofvessel extension. A possible explanation is that theprevious study measured extension along the direc-tion of the vascular arcades whereas the present mea-surements were taken along the parafoveal horizontalmeridian.
In summary, the precise laminar and regional ar-rangement of primate intrinsic retinal vessels arises bydifferent mechanisms of vascular development thatare partly segregated in space and time. As such, reti-nal vascularization provides an interesting model forangiogenesis research and suggests multiple avenuesof therapeutic intervention in diseases exhibiting ab-errant vascular growth.
The authors thank the Washington and Northern IdahoLions' Sight Conservation Foundation for human donor tis-sue. They also thank Andra Erickson for technical assistanceand the staff at the Tissue Distribution Program of the Re-gional Primate Research Center for their cooperation.
References
1. Snodderly DM, Weinhaus RS, Choi JC. Neural-vascu-lar relationships in central retina of Macaque monkeys(Macaca fascicularis). J Neurosci. 1992; 12:1169-1193.
2. Shimizu K, Ujiie K. In: Shimizu K, Ujiie K, eds. Struc-ture of Ocular Vessels. Tokyo: Igaku-Shoin Press; 1978.
3. Chan-Ling T, Stone J. Retinopathy of prematurity: Or-igins in the architecture of the retina. Prog Ret Res.1993; 12:155-178.
4. Foos RY. Retinopathy of prematurity: Pathologic cor-relation of clinical stages. Retina. 1987; 7:260-276.
5. Penn JS, Gay CA. Computerized digital image analysisof retinal vessel density: Application to normoxic andhyperoxic rearing of the newborn rat. Exp Eye Res.1992; 54:329-336.
6. Engerman RL. Development of the macular circula-tion. Invest Ophthalmol Vis Sci. 1976; 15:835-840.
7. Henkind P, Bellhorn RW, Murphy ME. Developmentof macular vessels in monkey and cat. BrJ Ophthalmol.1975; 59:703-709.
8. Ashton N. Retinal angiogenesis in the human embryo.BrMedBull. 1970;26:103-106.
9. Cogan DG. Development and senescence of the hu-man retinal vasculature. Trans Ophthalmol Soc UK.1963; 83:465-489.
10. Michaelson IC. The mode of development of the vas-cular system of the retina, with some observations onits significance for certain retinal diseases. Trans Oph-thalmol Soc UK. 1948;68:137-190.
11. Hendrickson A. A morphological comparison of fo-veal development in man and monkey. Eye.1992;6:136-144.
12. Yuodelis C, Hendrickson A. A qualitative analysis ofthe human fovea during development. Vision Res.1986; 26:847-855.
13. Lutty GA, McLeod DS. A new technique for visualiza-tion of the human retinal vasculature. Arch Ophthalmol.1992; 110:267-276.
14. Flower RW, McLeod DS, Lutty GA, Goldberg B, WajerSD. Postnatal retinal vascular development of thepuppy. Invest Ophthalmol Vis Sci. 1985; 26:9557-9568.
15. McLeod DS, Lutty GA, Wajer SD, Flower RW. Visual-ization of a developing vasculature. Microvasc Res.1987; 33:257-269.
16. Kretzer FL, Mehta RS, Goad D, Hittner HM. Animalmodels in research on retinopathy of prematurity. In:McPherson AR, Hittner HM, Kretzer FL, eds. Retinopa-thy of Prematurity. Toronto: BC Decker; 1986:79-88.
17. Hendrickson A, Kupfer C. The histogenesis of thefovea in the macaque monkey. Invest Ophthalmol.1976; 15:746-756.
18. Packer O, Hendrickson AE, Curcio CA. Develop-mental redistribution of photoreceptors across theMacaca nemestrina (Pigtail Macaque) retina. / CompNeurol. 1990; 298:472-493.
19. Patz A. The role of oxygen in retrolental fibroplasia.Pediatrics. 1957; 19:504-524.
20. Risau W, Lemmon V. Changes in the vascular extracel-lular matrix during embryonic vasculogenesis and an-giogenesis. Dev Biol. 1988; 125:441-450.
21. Chan-Ling T, Halasz P, Stone J. Development of reti-
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
Blood Vessel Development in Monkey Retina 3455
nal vasculature in the cat: Processes and mechanisms.CurrEyeRes. 1990;9:459-478.
22. Shakib M, De Oliveira LF, Henkind P. Developmentof retinal vessels: II: Earliest stages of vessel formation.Invest Ophthalmol Vis Sci. 1968; 7:689-700.
23. Huxlin KR, Sefton AJ, Furby JH. The origin and devel-opment of retinal astrocytes in the mouse. JNeurocytol.1992;21:530-544.
24. Gariano RF, Iruela-Arispe ML, Sage EH, HendricksonA. Development of Macaca monkey retinal vascula-ture. ARVO Abstracts. Invest Ophthalmol Vis Sci.1993; 34:898.
25. La Vail MM, Rapaport DH, Rakic P. Cytogenesis inthe monkey retina. / Comp Neurol. 1991;309:86-114.
26. Okada M, Erickson A, Hendrickson A. Light and elec-tron microscopic analysis of synaptic development inMacaca monkey retina as detected by immunocyto-chemical labeling for the synaptic vesicle protein, SV2.J Comp Neurol. 1994;339:535-558.
27. Stone J. The origins of the cells of vertebrate retina.Prog Ret Res. 1988; 7:1-19.
28. Connolly SE, Hores TA, Smith LEH, D'Amore PA.Characterization of vascular development in themouse retina. Microvasc Res. 1988; 36:275-290.
29. Kirby MA, Steineke TC. Morphogenesis of retinal gan-glion cells during formation of the fovea in the Rhesusmacaque. Vis Neurosci. 1992;9:603-616.
30. Ashton N, Cook C. Studies on developing retinal ves-sels: I: Influence of retinal detachment. BrJ Ophthal-mol. 1955; 39:449-455.
31. Buttery RG, Hendrichsen CFL, Weller WL, Haight JR.How thick should a retina be? A comparative studyof mammalian species with and without intraretinalvasculature. Vision Res. 1991;31:169-187.
32. Chase J. The evolution of vascularization in mammals.Ophthalmology. 1982;89:1518-1525.
33. Kageyama GH, Wong-Riley MTT. The histochemicallocalization of cytochrome oxidase in the retina andlateral geniculate nucleus of the ferret, cat, and mon-key, with particular reference to retinal mosaics andon/off-center visual channels. J Neurosci. 1984; 4:2445-2459.
34. Buttery RG, Weller, Haight JR. Retinal cytochromeoxidase: A missing link in the electron transportchain? Neurosci Lett. 1988;30(suppl):S52.
35. Lowry OH, Roberts NR, Lewis C. The quantitative his-tochemistry of the retina. J Biol Chem. 1956; 220:879-892.
36. Baird A, Esch F, Gospodarowicz D, Guillemin R. Ret-
ina- and eye-derived growth factors: Partial molecularcharacterization and identity with acidic and basic fi-broblast growth factor. Biochemistry. 1985; 24:7855-7860.
37. Chen C-H, Chen SC. Evidence for the presence ofa specific vascular endothelial growth factor in fetalbovine retina. Exp Cell Res. 1987; 169:287-295.
38. Hanneken A, Lutty GA, McLeod DS, Rovey F, HarveyAK, Hjelmeland LM. Localization of basic fibroblastgrowth factor to the developing capillaries of the bo-vine retina. / Cell Physiol. 1989; 138:115-120.
39. Breier G, Albrecht U, Sterrer S, Risau W. Expressionof vascular endothelial growth factor during angio-genesis and endothelial cell differentiation. Develop-ment. 1992; 114:521-532.
40. Plate KH, Breier G, Weich HA, Risau W. Vascular en-dothelial growth factor is a potential tumour angio-genesis factor in human gliomas in vivo. Nature.1992;359:845-848.
41. Shweiki D, Itin A, Soffer D, Keshet E. Vascular endo-thelial growth factor induced by hypoxia-initiated an-giogenesis. Nature. 1992;359:843-845.
42. Mclntosh LC, Muckersie L, Forrester JV. Retinal capil-lary endothelial cells prefer different substrates forgrowth and migration. Tissue Cell. 1988;20:193-209.
43. Roberts JM, Forrester JV. Factors affecting the migra-tion and growth of endothelial cells from microvesselsof bovine retina. Exp Eye Res. 1990;50:165-172.
44. Jiang B, Caldwell RB. Extracellular matrix changesand astrocyte migration during retinal vascular devel-opment in rats. ARVO Abstracts. Invest Ophthalmol VisSci. 1992;33:1089.
46. Frank RN. On the pathogenesis of diabetic retinopa-thy: A 1990 update. Ophthalmology. 1991; 98:586-593.
47. Plouet J, Chollet P, Moro F, Malecaze F. Secretion ofVAS/VEGF by retinal pericytes: A paracrine stimula-tion of endothelial cell proliferation. ARVO Abstracts.Invest Ophthalmol Vis Sci. 1993; 33:900.
48. Chan-Ling T, Stone J. Degeneration of astrocytes inretinopathy of prematurity causes failure of the blood-retinal barrier. Invest Ophthalmol Vis Sci. 1992; 33:2148-2159.
49. D'Amore P, Orlidge A, Herman I. Growth control inthe retinal microvasculature. Prog Ret Res. 1989;7:2233-2258.
Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933407/ on 02/07/2018
