Pericyte Changes in Branch Retinal Vein Occlusion

Ingolf H. L. Wallow, Colleen D. Bindley, Kathryn L. P. Linton, and Darius ftasregar

The model of experimental branch vein occlusion (BVO) in the monkey offers the opportunity to examine retinal capillaries under stress. Electron microscopic morphometry was done on 812 capillaries of 13 eyes of cynomolgus monkeys, comparing 579 capillary collaterals of 9 BVO eyes with 233 normal capillaries of 4 control eyes. The tissue underwent the myosin subfragment-1 technique to decorate and quantify bundles of actin filaments in capillary pericytes. The duration of BVO was 2-48 months. Capillary collaterals of BVO eyes had an enlarged caliber, endothelial hyperplasia, and pericyte hypertrophy, but no proportional increase in basement membrane area. Collaterals near the inner plexiform layer (IPL) had a greater wall thickness, pericyte coverage, and actin coverage than collaterals near the outer plexiform layer (OPL). Pericyte hypertrophy was proportionate to caliber increase in OPL vessels and exceeded caliber increase only in IPL vessels. Actin coverage was proportional with the vessel dilation and size of pericyte cytoplasm in all vessels. These findings indicate that capillary collaterals in BVO are not equipped morphologically for an increased regulatory role in microvascular flow beyond their normal function. Invest Ophthalmol Vis Sci 32:1455-1463,1991

For more than a century, pericytes have been sus- thus protected the blood vessel wall.14 In diabetes mel- pected of being contractile and involved in regulating litus where retinal capillaries undergo an early and capillary blood flow.1"4 The subsequent demonstra- profound loss of pericytes, vasoconstriction in re- tion in vascular pericytes of filamentous actin5"7 typi- sponse to 100% oxygen breathing was impaired.15 cal of smooth muscle cells,8 of myosin and of troporhy- Different organs have different pericyte coverage of osin,7-910 and of adhesion plaques linking pericytes vessels; for instance, coverage in skeletal muscle capil- and endothelial cells" confirmed that pericytes share laries is twice that in cardiac muscle. When both types biochemical similarities with smooth muscle cells and of capillaries were perfused separately with the vaso- are structurally specialized for contraction of the mi- constrictive agents angiotensin, norepinephrine, or crovascular wall. Moreover, retinal pericytes have vasopressin, comparative pressure increases in skele- been shown to contract in vitro during normal tal muscle capillaries were much higher.4 Tighter peri- growth12 or in response to Mg adenosine triphos- cyte coverage in skeletal microvessels was assumed to phate13 and to relax in response to the actin filament cause higher vasoconstrictive vascular resistance and disrupting agent, cytochalasin B.12 pressure increase. Conceivably, contraction of peri- In vivo indirect evidence for pericyte influence on cyte actin bundles in secondary (circumferential) retinal capillary blood flow was reported in experi- pericyte processes could narrpw vessel lumens, and mental systemic hypertension. Although brain capil- contraction within primary (longitudinal) processes laries broke down, retinal capillaries remained intact, could shorten the cellular extent and increase vascular permitting the interpretation that the comparatively tortuosity. Blood flow and vascular resistance would high number of contractile retinal pericytes may have be affected by both. counteracted the increase of hydrostatic pressure and Further indirect evidence for a role of retinal capillary pericytes in modulating blood flow may be col- lected from morphometry measuring the pericyte cov- From the Department of Ophthalmology, University of Wiscon- erage and content of actin filaments in pericytes. We sin, Madison, Wisconsin. recently refined an evaluation of the monkey model Supported by research grant EY-01634 from the National Eye of branch retinal vein occlusion1617 which produces Institute, National Institutes of Health, Bethesda, MD and by a grant from the Miller Foundation, Marshfield, Wisconsin (IHLW). brushes of capillary collaterals. These seem to form Presented in part at the Annual Meeting of the Association for under the "stress" of continued arteriolar inflow on a Research in Vision and Ophthalmology, May 4, 1990, Sarasota, compromised system of venous outflow. The density FL. of pericyte nuclei was unchanged in these vessels, but Submitted for publication: June 15, 1990; accepted October 24, cross sections showed redundancy of pericyte com- 1990. partments, suggesting a change in pericyte coverage. Reprint requests: Ingolf H. L. Wallow, MD, Department of Oph- thalmology, University of Wisconsin, Clinical Science Center, 600 With the hypothesis of a contractile role for retinal Highland Avenue, Madison, WI 53792. vascular pericytes receiving renewed interest, a suit-

Downloaded from iovs.arvojournals.org on 09/29/2021 1456 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / April 1991 Vol. 02

able model for microvascular stress being available, toluidine blue and correlated again with the fluores- and new techniques at hand to do ultrastructural mor- cein angiographic montages. When the full extent of a phometric measurements, we tried to determine if collateral brush was reached, a representative section pericytes might change adaptively with respect to was photographed, and the prints were mdntaged. their circumferential coverage and actin content in The collateral brush was identified on the montage, response to stress on their presumed modulatory role. and cross-sectioned small blood vessels next to or within the outer and inner plexiform layers (OPL and Materials and Methods IPL) were labeled and counted (Fig. 1). Brushes that contained at least five capiilary collaterals were cho- Animals sen for electron microscopic evaluation. Adjacent ul- In nine eyes of seven young adult female cynomol- trathin sections were cut, and stained with uranyl ace- gus monkeys we produced and evaluated retinal tate and lead citrate. All small blood vessels in a collat- branch vein occlusion (BVO) as described and illus- eral brush were examined in an electron microscope. trated elsewhere.16-17 Four fellow eyes of these mon- In control animals, capillaries from a corresponding keys served as normal controls. Attention for this retinal site were processed and photographed. study was focused on areas of collateral formation ("brushes") which were identified by fluorescein angi- Morphometric Techniques ography. The temporal raphe provided most collateral brushes although a few were also obtained from a Individual small blood vessels were photographed superior/inferior or nasal location. Monkeys were fol- at magnifications of 6000X in a single frame and at lowed for up to 4 years, and their eyes were enucleated 15,000X magnification for montaging. Calibration at 2, 3, 4, 10, 12, ISVi, 18, and 48 months after BVO. grids were photographed periodically, and the final The eyes were opened coronally through the pars magnification of a printed picture (approximately plana, allowing clear visualization under a dissecting 39,000x) was calculated individually from each nega- microscope of the posterior retina and correlation tive. The variation of final magnification in a sample with montages of the latest fluorescein angiogram, of 173 pictures was 1.4%. usually taken 1 week before euthanasia. Retinal capillaries may be defined by several parameters. However, even in the normal situation overlap in vessel size, transition of wall architecture, and simi- Tissue Processing larities of the cytoplasm make it difficult to distin- Retinal areas containing collateral brushes were guish pericytes of capillaries from the smooth muscle dissected in characteristic shapes for orientation and cells of precapillary arterioles. In pathologically al- glued onto squares of stainless-steel wire mesh, glycer- tered vessels, wall characteristics and dimensions may •V. inated, decorated with myosin subfragment-1 (S-l), change, eg, venous congestion may cause capillary di- fixed in a glutaraldehyde solution, and embedded in lation to the dimension of postcapillary venules. For an epoxy resin as previously detailed.1819 Light micro- our study, we defined capillaries by location (OPL or scopic step sections, 1.5-^m thick, were stained with IPL) and size (cross section fitting into one electron

*•

Fig. 1. Retinal cross-sections from monkey with experimental retinal branch vein occlusion showing small collaterals of collateral brushes next to or within the outer plexiform layer (arrows) and next to or within the inner plexiform layer (arrowheads). (A) Note excellent tissue preservation obtained by standard fixation. (B) Compromised preservation resulted from the glycerination step of the myosin subfragment-1 procedure used to identify actin filaments. A large "mature" collateral located within the inner retina is seen in the upper right hand corner (1.5 nm epoxy sections, toluidine blue, X280).

Downloaded from iovs.arvojournals.org on 09/29/2021 No. 5 PEPJCYTE CHANGES IN DVO / Wallow er ol 1457

micrograph of 3'A X 4-inch negative size at a magnifi- used so that all tracings and measurements were done cation of 6000X; this accommodated blood vessels by individuals having no knowledge of the exposure with outer diameters of up to 12 X 14.5 /im). If small groups involved. Quantitative planimetry was then blood vessels with other than capillary wall structure done with a model 2400 digitablet (Numonics, Lans- fit our selection criteria, they were assumed to be de- dale, CA) equipped with an electronic graphics calcu- rived from capillaries. We were unable to use the num- lator and interfaced with a Macintosh SE computer ber of endothelial cells as a parameter to identify capil- for capturing data. From each blood vessel, the follow- laries because the myosin S-l technique requires the ing primary measurements were obtained: fracturing of cell membranes by glycerination to allow the S-l to penetrate into the cell and decorate the 1. The IBM length (IBML) and outer (pericyte and actin filaments. Frequently, this actin identification endothelial combined) basement membrane technique caused sufficient artifactual damage to en- length (OBML). Correspondingly, the areas cir- dothelial cells (but not to basement membranes) so cumscribed gave the area occupied by the lumen that the endothelial cells and their cytoplasmic plus endothelial cell cytoplasm (IBMA) and the borders could not be specified. total area occupied by the blood vessel (OBMA), Only cross-sectioned vessels were evaluated further. with the difference between them indicating the Cross sections were defined by the ratio of maximum wall area (WA, the total space occupied by all base- to minimum diameter measured from the inside edge ment membranes plus all pericyte compartments). of the endothelial basement membrane across the lu- The sum of IBML and OBML divided by two indi- men. All vessels with a ratio equal to or greater than cated the average wall length, and the quotient of 2.0, vessels with significant technical artifacts of sec- WA divided by the average wall length gave the tioning, staining, or photographing, or vessels with average wall thickness (WT). the cross section of a pericyte nucleus were excluded. 2. Pericyte size (PS) and pericyte length (PL). The PS Using colored ink to delineate different aspects of indicated the area of each pericyte compartment the vessel, the envelopes and lengths of its wall com- containing viable or nonviable cytoplasm, and PL partments were traced. First, the internal outline of indicated the length of each compartment with via- the endothelial cell basement membrane and the ex- ble cytoplasm, as projected onto the IBM. ternal outline of the pericyte basement membrane 3. Actin area (A A) and actin length (AL). The A A were traced. Subsequently, each pericyte compart- indicated the size of each bundle of oriented actin ment and each bundle of parallel actin filaments ori- filaments, and AL the length of the bundle in the ented in longitudinal or oblique section were out- primary (ALP)/secondary/tertiary viable pericyte lined. Pericyte processes adjacent to endothelial cells processes. were termed "primary;" additional processes external From these primary measurements, other variables to the primary ones were termed "secondary/ter- were derived through statistical analysis. Ninety-eight tiary." All pericyte compartments and actin bundles blood vessels were remeasured to determine inter- were numbered. The vascular lumen was bisected by and intraobserver variability, using paired t-tests and the longest internal diameter. A second diameter per- coefficients of variation. Correlation coefficients and pendicular to the first was drawn through the mid- 95% confidence intervals were also calculated to as- point of the first. The midpoint of the second diame- sess inter- and intraobserver agreement. The variabil- ter was considered the center point of the vessel. Ra- ity ranged from 0.05-0.15%. In summary, we ob- dial rays were drawn from this center point to the tained and evaluated 812 blood vessels as specified in lateral borders of all pericyte compartments and all Table 1. actin filament bundles. The intersections of the rays with the inner (endothelial) basement membrane Statistical Evaluation (IBM) were used to measure the length of the pericyte processes and actin bundles. Pericyte compartments Initial examination of the data describing the wall containing only cytoplasmic debris were recorded as characteristics, namely IBML, OBML, IBMA, nonviable. The principles of our measurement tech- OBMA, WA, WT, PS, and AA, revealed a positive niques are illustrated in Figure 2 on a normal capil- skewness. A natural logarithmic transformation of lary and in Figures 3 and 4 on three collaterals located these measurements was used to correct this. The sta- at different levels in the neurosensory retina. tistical analyses reported used the transformed data. A Since the S-l technique induces tissue artifacts that significant interaction between treatment and retinal disrupt and obscure cell borders, every color-traced layer was detected by analysis of variance. Hence, all electron microscopic montage was checked individu- comparisons between experimental and control ves- ally by the principal investigator. A coding system was sels were done separately for each capillary location

Downloaded from iovs.arvojournals.org on 09/29/2021 1458 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / April 1991 Vol. 32

Fig. 2. Normal capillary cross section from monkey retina, labeled with myosin subfragment-1 technique to decorate actin filaments (A) within pericyte compartments (P) illustrates our measurement techniques as described in text. Note fragmentation of endothelial cytoplasmic membranes, but preservation of basement membranes and of actin filament bundles. Normal discontinuities of endothelial basement membranes toward pericytes were bridged by dashed lines indicating the presumed external border of endothelial cells (EM, x25,000).

group. The difference in the wall structure between ther IBMA or OBMA as parameters) and associated the experimental and control groups was assessed us- with greater WT, which in turn was correlated with an ing the student t-test. The Wisconsin Information enlarged area of cytoplasm within pericyte processes Storage and Retrieval, an information processing sys- (PS). Processes with both viable and nonviable cyto- tem, was used to process all data files. Statistical analy- plasm were measured and added. Nonviable cyto- ses were done using the SAS data analysis software plasm was found only among secondary and tertiary system (Cary, NC). processes. In primary pericyte processes, the length of The treatment and handling of al! animals in this bundles of circumferentially oriented actin filaments investigation conformed to the ARVO Resolution on (ALP) and the total area occupied by circumferen- the Use of Animals in Research. tially oriented actin filaments (AA) were greater in Results BVO collaterals than in control capillaries. Measure- ment of endothelial cytoplasm was excluded because Table 2 shows that, regardless of layer, the caliber of of substantial artifacts caused by the histochemical capillaries in BVO collaterals was enlarged (using ei- procedures for actin staining.

Downloaded from iovs.arvojournals.org on 09/29/2021 No. 5 PERICYTE CHANGES IN DVO / Wallow er ol 1459

Fig. 3. Two small blood vessels from collateral brush, (A from inner plexiform layer, B from outer plexiform layer). Tracing of capillary wall structures was omitted to allow original view of basement membranes and of actin filament bundles. Compared to Fig- ure 2, pericyte coverage is more extensive, and in- cludes segments of coverage

Jtt by two overlapping pericyte cytoplasmic processes (EM, XI 6,000).

r. \

Downloaded from iovs.arvojournals.org on 09/29/2021 1460 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / April 1991 Vol. 32

Fig. 4. Small blood vessel located within IPL of a collateral brush. This vessel was of larger caliber than vessels shown in Figures 2 and 3, but still within the definition given in text. Compartment outlines were copied from the original montage. Pericyte coverage is more extensive and wall composition more complex than in a normal blood vessel at this location (EM, XI 3,200).

Table 3 presents variables derived from our pri- region (Table 4). Actin coverage (ACP) along the mary measurements by statistical analysis. Pericyte IBML by actin bundles in the primary pericyte pro- coverage (PCP) along the IBML, directly opposite the cesses did not show significant differences between outer circumference of the endothelial cells, is a pro- BVO and control vessels of either region. portion characteristic of two length measurements ex- The total area of cytoplasm (PC A) occupied by pri- pressing the percentage of endothelial circumference mary, secondary, tertiary, and nonviable pericyte pro- covered by primary pericyte processes. Normal PCP cesses relative to the overall wall area was approxi- ranged from 71-74%. It was larger for BVO of the IPL mately 47-48% in normal vessels. In BVO vessels of

Downloaded from iovs.arvojournals.org on 09/29/2021 No. 5 PERICYTE CHANGES IN BVO / Wallow er ol 1461

Table 1. Small blood vessels from nine BVO and as a "contractile index" relating the potentially avail- four control eyes evaluated by computerized able actin force to the blood vessel wall area that needs planimetry and statistical analysis to be "toned." Both parameters (AVP and ACA) showed no significant differences between BVO and Number of small blood vessels from control vessels. BVO eyes Control eyes Normal capillaries from the IPL and OPL areas showed no significant differences in their wall charac- Total OPL\ IPL Total OPL IPL Age* (mo) teristics and derived parameters, but capillary collat- 2 117 14 103 erals in BVO had a number of significant area effects 2 50 13 37 (Table 4); IPL vessels showed greater wall thickening 3 37 24 13 4 28 22 6 (WA and WT) than OPL vessels (Table 2) and had 10 111 59 52 94 48 46 more viable pericyte processes and actin bundles (Ta- 12 49 24 25 25 12 13 ble 5). The amount in overall pericyte cytoplasm and 15'/2 12 7 5 22 11 11 18 107 72 35 92 53 39 overall actin was also higher in IPL vessels (apparent 48 68 32 36 in the transformed data). Furthermore, the endorse- Total 579 267 312 233 124 109 ment of the vessel walls in the IPL area was greater in terms of PCP, PCA, and ACP. OPL = outer plexiform layer; IPL = inner plexiform layer. * Duration after complete BVO. t Retinal layer near which blood vessels were sampled. Discussion The caliber of capillary collaterals developing as a the IPL region, it was significantly higher, approxi- result of retinal BVO in areas of collateral brushes was mately 53%. This difference was less significant when larger than that of normal capillaries in non-BVO nonviable pericyte processes were excluded. eyes. In a prior light microscopic study using trypsin- The PCA also expresses the proportion of wall digested preparations and measurements of blood thickness contributed separately by pericyte cyto- vessel cross sections, we found an increase in mean plasm versus all basement membranes. We calculated vessel caliber of 44% (from 7.2 ± 0.8 /urn for control directly the proportional area occupied by all base- vessels to 10.4 ± 2.5 /urn for BVO capillary collaterals; ment membranes (BMA) in the vessel wall with the P < 0.01).16 In that study measurements were taken formula: (WA - viable PS - nonviable PS)/WA. For outside the vicinity of endothelial and pericyte nuclei IPL vessels BMA was significantly smaller. but included the thickness of the vessel wall, ie, endo- The combined area of actin filament bundles in all thelial cytoplasm, pericyte cytoplasm, and associated pericytes in relation to the cytoplasmic area of all peri- basement membranes. Also, vessels from both areas, cytes (AVP) is the potentially contractile part of peri- within/near the IPL and within/near the OPL were cytes in proportion to the size of cytoplasm that lumped together. would be affected by actin filament contraction. Simi- In our present electron microscopic study, the true larly, ACA refers to actin coverage, giving the overall caliber was defined more accurately due to different area of actin filaments in all pericyte processes in pro- fixation and processing (glutaraldehyde and embed- portion to the overall wall area. It may be thought of ding in an epoxy resin) and the higher resolution of

Table 2. Wall characteristics of control vessels and BVO collaterals (mean ± SD) of monkey retina at the level of inner plexiform layer (IPL) and outer plexiform layer (OPL)

IPL Area OPL Area

Control BVO Control BVO (n == 109) (n>. = 312) % Increase (n = 124) (n =•267) % Increase

2 OBMA (Mm ) 32 (li) 49 (23) 53 32 (11) 50 (29) 56 2 IBMA (Mm ) 19 (7) 30 (16) 58 20 (7) 32 (24) 60 2 WA (Mm ) 13 (6) 19 (9) 46 12 (5) 17 (8) 42 OBML(Mm) 21.3 (3.7) 26.8 (6.4) 26 21.0 (3.8) 26.6 (7.7) 27 IBML(Mm) 16.7 (3.0) 20.5 (5.3) 23 16.9 (3.2) 20.9 (6.8) 24 WT (Mm) 0.65 (0.24) 0.78 (0.25) 20 0.62 (0.20) 0.72 (0.22) 16 2 PS (Mm ) 6 (3) 10 (6) 67 6 (3) 8 (5) 33 ALP (Mm) 8.3 (3.4) 10.7 (3.9) 29 7.7 (2.8) 9.7 (4.8) 26 2 AA (Mm ) 2 (1) 2 (1) 1 (1) 2 (1) For all parameters listed, the /"-value was <0.0001. Statistical evaluation was based on natural logarithmic transformation of measurements.

Downloaded from iovs.arvojournals.org on 09/29/2021 1462 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / April 1991 Vol. 02

Table 3. Derived wall parameters of control vessels and BVO collaterals in monkey retina (mean ± SD)

IPL OPL

Control BVO Control BVO (n = 109) (n = 312) P'-value (n = 124) (n = 267) P'-value

PCP% 73.6(13.8) 78.4(13.7) 0.0020 71.0(13.7) 73.6(17.9) 0.1085 ACP% 48.9(17.5) 52.1 (14.6) 0.0888 45.2(13.1) 46.6(17.7) 0.3527 PCA% 47.9 (7.0) 52.8(11.3) 0.0001 46.8 (9.4) 46.9(10.9) 0.9485 PCA%* 47.8(7.1) 50.6(12.7) 0.0051 46.6 (9.6) 45.0(11.6) 0.1707 BMA% 51.6(7.0) 46.8(11.2) 0.0001 52.7 (9.3) 52.8(10.8) 0.9495 AVP% 25.7(7.5) 24.1(8.7) 0.0809 25.0(7.9) 25.3 (9.6) 0.7323 ACA% 11.1 (3.4) 11.7(4.1) 0.1753 10.3(3.1) 10.9 (4.6) 0.1487

The parameters PCP, ACP, PCA, AVP, and ACA are explained in the text. Value of PCA when only viable pericytes were counted. IPL/OPL = area of inner/outer plexiform layer.

the method. We measured the IBMA which included sociated with structural changes of the vessel wall. the lumen and the endothelial cytoplasm but ex- First, WT in all capillary collaterals increased, particu- cluded the rest of the vessel wall. Overall values for larly in collaterals of the IPL region. The thickening this parameter increased by 63% (from 19 ± 7 fim2 to was due to expansion of pericyte cytoplasm. In IPL 31 ± 20 ^m2; P < 0.0001). When the rest of the vessel collaterals, the pericyte coverage appeared more ex- wall was included (OBMA), overall values increased tensive, and the number of viable pericyte processes by 53% (from 32 ± 11 /urn2 to 49 ± 26 ^m2; P higher. Interestingly, WT was not correlated with a < 0.0001). Furthermore, IPL and OPL vessels were disproportionate thickening of the overall BMA as is measured separately, and we found that both partici- typical for other chronic microvascular conditions of pated to almost the same degree in the caliber increase the retina, eg, experimental hypergalactosemia.23 In- associated with BVO (Table 2). stead, BVO collaterals showed no increase of their rel- That collateral brushes in BVO develop from the ative overall basement membrane contribution. In existing capillary framework and represent dilated normal capillaries, the pericyte cytoplasm contrib- capillaries is a well-known clinical20 and experimen- uted 47-48% to the WA, and the basement mem- tal21 observation. However, formal measurement of branes contributed 52-53%. In OPL collaterals, this small collateral channels has not been detailed before ratio was maintained. In IPL collaterals, it was re- 16 to our knowledge. Also, in the past, collaterals in gen- versed. In our previous study we found a stable num- eral were said to occupy the "inner retinal layers."22 ber of pericyte nuclei. The current data of expansion This seems to be true for large-caliber preferential of pericyte cytoplasm suggest widespread pericyte hy- channels developing in collateral brushes, but small pertrophy, proportionate with the caliber increase in collaterals in our study remained localized at the in- OPL collaterals and in excess of caliber increase in termediate and deep capillary beds. IPL collaterals. Capillary dilation to form collateral brushes was as- The second structural change of collateral walls involved the cytoplasmic content of filamentous actin in the pericytes. The total circumferential length of Table 4. Area effect comparing normal vessels and actin bundles in the primary processes of BVO collat- BVO collaterals in two different layers of monkey erals increased significantly but commensurate with retina (IPL vs. OPL region) vessel dilation (ie, the proportionate length of actin Wall characteristics bundles did not increase over that of control vessels). Similarly, the pericyte area occupied by actin was Control BVO larger in BVO collaterals than in control vessels. The t-statistic ¥-value (-statistic P- value OPL vessels which did not seem to form additional pericyte processes (Table 5) had a larger increase than WA 0.8253 0.4104 2.6872 0.0074 IPL vessels whose pericytes seemed to expand into WT 1.0190 0.3093 3.6556 0.0003 PS 1.2507 0.2123 4.8349 0.0001 additional circumferential processes, but the propor- AA 1.6463 0.1011 4.0114 0.0001 tionate area of actin filaments relative to the size of the pericyte cytoplasm remained the same in both re- Derived parameters gions. PCP 1.4663 0.1439 3.5587 0.0004 Thus, BVO collaterals from collateral brushes are ACP 1.8288 0.0689 4.0010 0.0001 dilated capillaries with endothelial cell hyperplasia16 PCA 0.9699 0.3331 6.4692 0.0001 and pericyte hypertrophy producing a thickened ves-

Downloaded from iovs.arvojournals.org on 09/29/2021 No. 5 PERICYTE CHANGES IN DVO / Wallow er al 1463

Table 5. Summary of number of pericyte compartments and actin bundles (mean ± SD) IPL OPL

Control BVO P-value Control BVO P-value

No. viable pericytes 5.9 (2.2) 6.8(3.1) 0.0054 5.9 (2.5) 6.1(2.9) 0.5084 No. nonviable pericytes 1.0(0.0) 1.0(1.1) 1.000 1.0(0.3) 1.2(1.1) 0.0474 No. actin bundles 9.0(4.1) 11.5(5.2) <0.0001 8.9 (3.9) 9.9 (4.8) 0.432

sel wall. If collateral formation is interpreted correctly 8. Skalli O, Pelte MF, Peclet MC, Gabbiani G, Gugliotta P, Bus- as a response to the hemodynamic stress of BVO, the solati G, Ravazzola M, and Orci L: a-smooth muscle actin, a adaptive wall thickening seems more vigorous in IPL differentiation marker of smooth muscle cells, is present in micronlamentous bundles of pericytes. J Histochem Cytochem collaterals. Closer proximity to the arteriolar side of 37:315, 1989. the vascular tree may support this response. The OPL 9. Joyce NC, Haire MF, and Palade GE: Contractile proteins in collaterals (believed to be more venous in character) pericytes: I. Immunoperoxidase localization of tropomyosin. J participated in compensatory changes, and at least in Cell Biol 100:1379, 1985a. the otherwise healthy monkey retina, did not develop 10. Joyce NC, Haire MF, and Palade GE: Contractile proteins in pericytes: II. Immunocytochemical evidence for the presence the morphologic picture of disabled passive conduits. of two isomyosins in graded concentrations. J Cell Biol Interestingly, pericyte hypertrophy included an in- 100:1387, 1985b. crease of filamentous actin but in proportion to the 11. Courtoy PJ and Boyles J: Fibronectin in the microvasculature: caliber increase of the stressed vessels. The absence of Localization in the pericyte-endothelial interstitium. J Ultra- a disproportionate increase of actin may indicate that struct Mol Struct Res 83:258, 1983. collaterals from collateral brushes in retinal BVO are 12. Kelley C, D'Amore P, Hechtmann HB, and Shepro D: Micro- vascular pericyte contractility in vitro: Comparison with other not equipped for an unusual regulatory role in micro- cells of the vascular wall. J Cell Biol 104:483, 1987. vascular flow beyond their normal function. Other 13. Das A, Frank RN, Weber ML, Kennedy A, Reidy C, and Man- stresses like prolonged systemic hypertension may be cini MA: ATP causes retinal pericytes to contract in vitro. Exp more effective in provoking disproportionate actin in- Eye Res 46:349, 1988. 14. Laties AM, Rapaport SI, and McGlinn A: Hypertensive break- creases. Corresponding experiments and evaluations down of cerebral but not of retinal vessels in rhesus monkey. in this laboratory are underway. Arch Ophthalmol 97:1511, 1979. 15. Grunwald JE, Riva CE, Brucker AJ, Sinclair SH, and Petrig Key words: pericytes, actin filaments, branch retinal vein BL: Altered retinal vascular response to 100% oxygen breath- occlusion, capillary collaterals, cynomolgus monkey ing in diabetes mellitus. Ophthalmology 91:1447, 1984. 16. Danis RP and Wallow IH: Microvascular changes in experimental branch retinal vein occlusion. Ophthalmology 94:1213, References 1987. 17. Wallow IH, Danis RP, Bindley C, and'Neider M: Cystoid macu- 1. Rouget C: Memoire sur le developpement, la structure et les lar degeneration in experimental branch retinal vein occlusion. proprietes physiologiques des capillaries sanguins et lymphati- Ophthalmology 95:1371, 1988. ques. Archives de Physiologie Normale et Pathologigue 5:603, 18. Wallow IH, Greaser ML, and Stevens T: Actin filaments in 1873. diabetic fibrovascular membrane. Arch Ophthalmol 99:2175, 2. Zimmerman K.W: Der feinere Bau der Blutcapillaren. 1981. Zeitschrift fur die gesamte Anatomie 68:29, 1923. 19. Wallow IH, Stevens TS, Greaser ML, Bindley C, and Wilson R: 3. Rhodin J: The ultrastructure of mammalian arterioles and pre- Actin filaments in contracting preretinal membranes. Arch capillary sphincters. J Ultrastruct Mol Struct Res 18:181, 1967. Ophthalmol 102:1370, 1984. 4. Tilton RG, Kilo C, Williamson JR, and Murch DW: Differ- 20. Wise GN, Dollery CT, and Henkind P: The Retinal Circula- ences in pericyte contractile function in rat cardiac and skeletal tion. New York, Harper & Row, 1971, pp. 352-358. muscle microvasculatures. Microvasc Res 18:336, 1979. 21. Kohner EM, Dollery CT, Shakib M, Henkind P, Waterson JW, 5. Wallow IH and Burnside B: Actin filaments in retinal pericytes de Oliveira NF, and Bulpitt CJ: Experimental branch vein oc- and endothelial cells. Invest Ophthalmol Vis Sci 19:1433, clusion. Am J Ophthalmol 69:778, 1970. 1980. 22. Henkind P and Wise GN: Retinal neovascularization, collat- 6. Herman IM and D'Amore PA: Microvascular pericytes con- erals, and vascular shunts. Br J Ophthalmol 58:413, 1974. tain muscle and nonmuscle actins. J Cell Biol 101:43, 1985. 23. Robison WG, Kador PF, and Kinoshita JH: Retinal capillaries: 7. Gordon SR and Essner E: Actin myosin, and laminin localiza- Basement membrane thickening by galactosemia prevented tion in retinal vessels of the rat. Cell Tissue Res 244:583, 1986. with aldose reductase inhibitor. Science 221 (4616): 1177, 1983.

Downloaded from iovs.arvojournals.org on 09/29/2021