Systemic Hypertension Produces Pericyte Changes in Retinal

Ingolf H. L. Wallow,* Colleen D. Bindley,* David M. Reboussin,f Stephen J. Gange,f and Marian R. Fishery-

Purpose. The purpose of this study was to examine retinal capillaries and their pericytes that previous research suggests to be contractile. A contractile role regulating blood flow may be more apparent when the vasculature is subjected to the stress of systemic hypertension. Methods. Using ultrastructural morphometry and the myosin subfragment-1 technique, retinal capillaries of normal and hypertensive rats were measured at three different time points, early, intermediate, and late (24, 44, and 68 wk). Results. Hypertensive capillaries seemed to dilate at the early time point (P = 0.002), were constricted at the intermediate time point (P < 0.001), and did not redilate later. Wall thick- ness was enlarged at all times, pericyte coverage (the ratio of plasma membrane length in contact with the vascular circumference to the outer circumference of the endothelial tube) was greater at early and intermediate time points, and the total area of viable cytoplasm relative to the vessel wall area was increased at the intermediate time (all P < 0.001). Also, at the intermediate time, the circumferential coverage of the endothelial tube by filamentbun - dles within pericytes and the actin area relative to the vessel wall area had increased (P < 0.001). Conclusions. These data indicate that the effects of systemic hypertension extend into the retinal capillary bed, causing pericyte change with actin increase and capillary constriction. They represent the first in vivo indirect evidence by morphologic criteria for pericyte contrac- tility in retinal vascular disease. Invest Ophthalmol Vis Sci. 1993; 34:420-430.

XVetinal capillary pericyte architecture suggests that associated. They possess suitably arranged contractile pericytes are -like cells that may modu- actin filaments1 and direct pericyte-endothelial cell cy- late the caliber of the blood vessels with which they are toplasmic contacts that may be related to the transmis- sion of nonnervous communication between contrac- tile pericytes and the endothelial cell cylinder.1'2 In From the Departments of *Ophthalmology and -fliiostatistics, University culture, retinal capillary pericytes synthesize the of Wisconsin, Madison, Wisconsin. potent vasodilator prostacyclin,3 and reportedly have Presented in part at the Annual Meeting of the Association for Research functional receptors for autonomic neurotransmit- in Vision and Ophthalmology, Sarasota, Florida, May, 1992. Supported by research grant EY-01634 from the National Eye Institute, ters, suggesting that pericytes might respond to auto- National institutes of Health, Bethesda, Maryland, by a grant from the nomic vasoactive substances in vivo.4 Miller Foundation, Marshfield, Wisconsin (IHLW), and by an If pericytes play a role in modulating retinal capil- •unrestricted grant from Research to Prevent Blindness, Inc., New York, to the Department of Ophthalmology. lary blood flow, one would expect an adaptive change Submitted for publication: May 22, 1992; accepted September 11, 1992. during chronic "stress" of the microvasculature. In- Proprietary interest category: N. deed, in experimental retinal branch occlusion Reprint requests: Ingolf H. L. Wallow, Department of Ophthalmology, University of Wisconsin, Clinical Science Center, 600 Highland Avenue, where a compromised outflow exposes portions of the Madison, WI 53792. capillary bed to continued arteriolar pressure, af-

Investigative Ophthalmology & Visual Science, February 1993, Vol. 34, No. 2 420 Copyright © Association for Research in Vision and Ophthalmology

Downloaded from iovs.arvojournals.org on 09/27/2021 Retinal Pericytes in Hypertension 421

fected vessels showed caliber increase and pericyte hy- Tissue Processing pertrophy. This hypertrophy included an increase in Before being killed, the rats were anesthetized and the filamentous actin proportionate to the caliber in- 5 eyes enucleated. From the horizontal meridian adja- crease. Systemic hypertension is another disease that cent to the temporal margin of the optic nerve head in may involve retinal capillaries. Degenerative changes each eye, a 3 X 5 mm strip of retina and choroid was of pericytes alone or of pericytes and endothelial cells dissected. To maintain orientation, the tissue was as well as rare indications for increased pericyte vol- placed on a Teflon sheet and then glued onto a square ume were found in monkeys with several months of 5 6 of stainless steel wire mesh, glycerinated, decorated systemic blood pressure elevation and in rats of short- with myosin subfragment-1, fixed in a glutaraldehyde term hypertension where inconstant changes in retinal 7 solution, and embedded in an epoxy resin, as previ- capillary pericytes were described. We assumed that ously detailed.8 Step sections, 1.5 /im thick, were morphometric techniques used in the branch vein oc- stained with toluidine blue and examined under a light clusion study cited above might reveal such changes in microscope for the presence of capillaries within both greater detail and accuracy. In particular, we won- the outer and the inner plexiform layers (OPL, 1.PL), dered if the generalized sustained stress of systemic and for the concurrent absence of large vessels in the hypertension also might produce pericyte hyper- other layers of the retina. When such an area was trophy with an actin increase exceeding any simulta- found, the section was photographed at X50 magnifi- neously occurring caliber change. Such an adaptive cation (Fig. 1) and adjacent ultrathin sections were change might add to the body of evidence that retinal cut, stained with uranyl acetate and lead citrate, and capillary pericytes serve a caliber-modulating function examined in an electron microscope. The montaged and thus influence microvascular circulation in health photographs of the semithick sections were used to and disease. identify the exact location of each capillary examined.

MATERIALS AND METHODS Animals • 4 Spontaneously hypertensive male rats (SHR) and non- hypertensive male control rats (WKY) were entered into this study. The 8-wk-old rats were randomly as- signed to three groups: group 1, short-term or "early" (24 wk); group 2, intermediate-term (44 wk); and group 3, long-term or "late" (68 wk). Group 1 had 30 animals each in the SHR and WKY categories, group 2 had 29 SHR and 30 WKY rats, and group 3 had 22 and 12 animals, respectively. The animals were group- housed (five to six per cage), maintained on a NIH 31 general diet with water ad libitum, and habituated to measurement procedures during several training ses- sions. At regular intervals the animals were weighed, and under the same experimental conditions the systolic :*L blood pressure was determined using a tail cuff, pro- grammed electrosphygmomanometer, photoelectric flow sensor, amplifier, and recorder (system by IITC, Woodland, CA). During measurement sessions, the rats were kept at room temperature of 22°C and not tranquilized to avoid blood pressure depression. Peak arterial pulse oscillations from two to five cuff defla- tions were obtained and averaged to give one result per measurement session. Initial systolic blood pres- FIGURE l. Retinal cross-sections from SHR rat at interme- sures were ] 81 ± 17 (SHR) versus 131 ± 12 (WKY). A diate time point showing- capillaries adjacent to inner and outer plexiform layers (arrows). (A) Note excellent, tissue statistically significant separation of blood pressure preservation obtained by standard fixation. (B) Compro- values between the experimental and the control ani- mised preservation resulted from the glycerination slep of mals was maintained throughout the duration of the the myosin subfragment-1 procedure used to identify actin study. filaments. (1.5 jLtm epoxy sections, toluidine blue, X325).

Downloaded from iovs.arvojournals.org on 09/27/2021 422 Investigative Ophthalmology & Visual Science, February 1993, Vol. 34, No. 2

The handling of animals in this investigation con- calibration of the enlarger was checked against every formed to the ARVO Resolution on the Use of Ani- fifth montage printed. Variability between enlarge- mals in Research. ments from vessels of a given animal were so small that the mean value was applied to all vessels of that animal. Morphometric Techniques Using montages, linear and area measurements A previously detailed protocol was used.5 Accordingly, (see below) were obtained from each vessel through capillaries were defined by location (OPL or IPL) and computerized planimetry by individuals having no size (cross-section fitting into one electron micrograph knowledge of the experimental groups involved. Every at primary magnification of X6000), and only cross- color-traced electron microscopic montage was sectioned capillaries were evaluated further. Electron checked individually by the principal investigator, micrographs from 9 to 24 capillaries per animal were again without knowledge of experimental groups. An obtained, 4 to 14 from OPL and 3 to 10 from IPL. example of a montage prepared for planimetry is Each animal was represented by capillaries of one eye given in Figure 2. Each of the vessel measurements only. were then standardized by dividing all linear measure- Electron micrographs were enlarged approxi- ments by the total magnification and all area measure- mately 2.6 times to print montages of each capillary. ments by the squared magnification. Calibration of the electron microscope was performed From each , the following primary at least once during examination of each animal, and measurements were obtained:

FIGURE 2. Capillary cross-section from late time point SHR rat illustrates measurement tech- niques described in text. Section is labeled with myosin subfragment-1 technique to decorate actin filaments (A) within pericyte compartments (P). Note fragmentation of endothelial cytoplasmic membranes, but preservation of basement membranes (OBM, IBM) and of actin filament bundles. (Electron microscopy, original magnification X29.500.)

Downloaded from iovs.arvojournals.org on 09/27/2021 Retinal Pericytes in Hypertension 423

1. The inner length (IBML) lowing variables: IBMA, OBMA, WA, WT, AA, PCP, and outer (pericyte and endothelial cell com- ACP, VPCA, AVP, ACA. bined) basement membrane length (OBML). A repeated-measures analysis of variance model Correspondingly, the areas circumscribed gave was used to distinguish the variability between rats (in- the area occupied by the lumen plus endothelial teranimal variability) and between retinal layers within cell cytoplasm (the inner basement membrane rats (intra-animal variability). Differences between the area, IBMA) and the total area occupied by the two experimental groups (SHR and WKY) and differ- blood vessel (outer basement membrane area, ences between the three age groups (24, 44, and 68 OBMA). The difference between them indicated wk) were assessed relative to the interanimal variability the wall area (WA, the total space occupied by all using measurements averaged over the two retinal basement membranes plus all pericyte compart- layers. Differences between the retinal layers were as- ments). The sum of IBML and OBML divided by sessed relative to the intra-animal variability. F-tests two indicated the average wall length, and the were used to investigate the significance of differ- quotient of WA divided by the average wall ences. Tests of overall interanimal and intra-animal length gave the average wall thickness (WT). differences were performed at a significance level of 2. Pericyte area (PA) and pericyte length (PL). PA 0.001 to protect against spurious relationships arising indicated the area of each pericyte compartment from the large number of variables examined. For sig- containing viable (VPA) or nonviable cytoplasm nificant overall differences, additional F-tests of spe- (NVPA). We designated pericyte cytoplasm as cific effects were tested at a level of 0.001. nonviable when it consisted of debris without rec- The total number of vessels included in this statis- ognizable organelles. Pericyte processes adja- tical evaluation was 963 for SHR (506 in OPL, 457 in cent to endothelial cells were termed "primary" IPL) and 871 for WKY (441 for OPL, 430 for IPL). and were always viable. Additional pericyte pro- The respective numbers for the three age groups were cesses external to the primary ones (ie, "second- at 24 wk, 358 (198/160) and 382 (194/188); at 44 wk, ary/tertiary" processes), however, on a few occa- 344 (174/170) and 347 (175/172); at 68 wk, 261 sions contained nonviable cytoplasm in both (134/127) and 142 (72/70). The frequency distribu- WKY and SHR animals (see Results). PL indi- tion of the number of vessels for both the inner and cated the length of each compartment with via- outer plexiform layers was similar between WKY and ble cytoplasm, as projected onto the inner base- SHR rats at each time point. ment membrane (IBM), PLP was the PL within primary pericyte processes. RESULTS 3. Actin area (AA) and actin length (AL). The AA indicated the size of each bundle of oriented ac- None of the parameters of the capillary wall differed tin filaments, and AL the length of the bundle in significantly between OPL and IPL within each group primary (ALP)/secondary/tertiary viable peri- of animals in any age group. Therefore, OPL and IPL cyte processes. were combined. For caliber evaluation, the parameter IBMA was From these primary measurements, other vari- used. Significant age-dependent caliber changes oc- ables were derived through statistical analysis. curred among and between hypertensive and normo- 1. PCP = Pericyte coverage (of primary pericyte tensive vessels (Table 1 and Fig. 3). Initially, at 24 wk, processes) as a percentage of inner basement SHR vessels seemed significantly larger than WKY ves- length, equal to PLP/IBML. sels (P = 0.002). By 44 wk, however, the caliber of 2. ACP = Actin coverage (of primary pericyte pro- SHR vessels had constricted (P < 0.001) and was no cesses) as a percentage of inner basement length, longer different from normotensive WKY capillaries equal to ALP/IBML. (P = 0.785). Later, at 68 wk, SHR vessels had retained 3. VPCA = Total area of viable pericyte cytoplasm a caliber significantly smaller than at the beginning (P relative to overall wall area, equal to VPA/WA. = 0.009), and had not significantly changed from 44 4. AVP = Actin coverage as a percentage of viable wk (P = 0.242). Normal vessels that had retained the pericyte area, equal to AA/VPA. same caliber between 24 and 44 wk (P - 0.449) 5. AC A = Actin coverage as a percentage of wall showed at 68 wk a significant dilation compared to 24 area, equal to AA/WA. wk (P = 0.002), compared to 44 wk (P < 0.001), and compared to hypertensive capillaries at 68 wk (P Statistical Evaluation = 0.006). The overall change with time between 24 For each primary measurement and derived variable, and 68 wk was a significant caliber decrease in hyper- the average for each retinal layer (OPL and IPL) in tensive vessels (P = 0.009) and a significant increase in each animal was calculated. Models were fit to the fol- the normotensive vessels (P = 0.002). This "cross-

Downloaded from iovs.arvojournals.org on 09/27/2021 424 Investigative Ophthalmology 8c Visual Science, February 1993, Vol. 34, No. 2

TABLE l. Caliber of Retinal Capillaries in and Fig. 3). In all three age groups, the wall thickness SHR and WKY Rats* (WT) was greater in hypertensive than in control ves- sels (P = 0.001; P < 0.001; P = 0.001, for the three age Inner Basement Membrane Area (nm2) Age groups, respectively). In comparing 24 with 68 wk, (wk) WKY SHR there was, moreover, a significant age-dependent in- crease of WT among hypertensive (P < 0.001), but not 1 2 2 3 4 24 13.07 (30) - 14.50 (30) - - among normotensive vessels (P = 0.02). 44 12.72 (30)5 12.59 (29)3 4 6 Within the vessel wall there was a change in the 68 14.98 (12)1-6-6 13.18 (22) - pericytes. The total length of primary pericyte pro- * Caliber expressed as inner basement membrane area, with each cesses (PLP) and of circumferentially oriented actin rat represented by one eye. Number of eyes noted in parentheses. filament bundles (ALP) within them, as well as their 1,2: P= .002. 8,5: P < .001. total actin area, all were significantly greater in SHR 4: /J= .009. than in WKY capillaries (Table 2). There was no differ- 6: P = .006. ence among capillaries attributable to layer (OPL vs. IPL). Nonviable pericyte processes were slightly more frequent in SHR vessels (4.5% vs 3.2%), but not signifi- over" relationship is detailed in Table 1 and dia- cantly so. They were found only among secondary and grammed in Figure 3. One needs to remember that in tertiary processes (ie, external to the primary pro- the oldest age group (68 wk), the sample size was re- cesses that lined the endothelial basement membrane). duced, accounting in part for the reduced P-value. Derived variables are also presented in Table 2 Wall changes between hypertensive and normo- and in Figures 4 and 5. Pericyte coverage (PCP) along tensive retinal capillaries also were significant (Table 2 the IBML was significantly increased in SHR capillar-

Caliber (by IBMA) - u2 Wall thickness -

C\J d

oo CO

CO CO d

o CO O 20 30 40 50 60 70 20 30 40 50 60 70 Age in Weeks Age in Weeks FIGURE 3. Age-dependent and hypertension-dependent changes were seen in caliber (IBMA, inner basement area) and wall thickness of SHR (1) and WKY (0) vessels. For details, see text. Vertical bars indicate standard error of the mean. Values for the SHR rats sacrificed at each lime point are connected by a solid line and for WKY rats by a dashed line to emphasize the two distinct types of animals. Data are not available other than at the points of sacrifice, and assumptions about the pattern of response in the intervals should not be inferred by the connecting lines.

Downloaded from iovs.arvojournals.org on 09/27/2021 Retinal Pericytes in Hypertension 425

TABLE 2. Wall Characteristics of Retinal Capillaries in SHR and WKY Rats* Mean

WKY SHR Standard Relative Wall Characteristics (n = 72) (n = 81) Error^ Change (%)% P Valued

Wall Thickness (/jin) WT 3.32 3.85 0.25 16.0 <.001 Area (M>II2) WA 4.96 5.79 3.86 16.7 <.001 Basement membrane Inner length (/urn) IBML 13.78 13.73 1.34 -0.7 .79 Outer length (/xm) OBML 15.73 15.99 1.52 1.9 .15 Inner area (fim2) IBMA 13.59 13.42 2.51 -1.2 .62 Outer area (MHI2) OBMA 18.54 19.21 3.39 3.6 .10 Pericyte Length (urn) PLP 5.63 6.68 1.69 18.6 <.001 %of IBML PCP 41 48 29 17.1 <.001 Area (MIH2) VPA 2.29 2.89 0.95 26.2 <.001 % of WA VPCA 44 48 24 9.1 <.001 Actin Length (/urn) ALP 3.53 4.48 1.47 26.9 <.001 %of IBML ACP 26 32 28 23.1 <.001 Area (^m2) AA 0.67 0.94 0.38 39.7 <.001 % of VPA AVP 30 33 13 10.0 .25 % of WA ACA 13 16 11 23.1 <.001 * See morphometric techniques section of "Materials and Methods" for detailed description of wall characteristics. f A pooled standard error estimate is given because the variability between the two types of animals was comparable. % Relative change calculated as mean SHR minus mean WKY divided by mean WKY. § The /•" value is given for the statistical comparison of the mean WKY estimate versus the mean SHR estimate (see statistical evaluation section of "Materials and Methods"). Differences were considered important if the P value was <.001, a technique for multiple compari- sons using the same data set.

ies over WKY vessels (P < 0.001 at 24 and 44 wk, P (Fig. 5). Earlier, at 24 wk, and later, at 68 wk, P values = 0.004 at 68 wk). Again, fewer observations in the were 0.04 and 0.62, respectively. An age-dependent long-duration group may account in part for the re- decrease of VPCA also was noted. In hypertensive ves- duced P value, even though the difference between sels, this occurred late (P = 0.002 between interme- SHR and WKY vessels at 68 wk remained the same as diate and late, P < 0.001 between early and late); in at the earlier stages (Fig. 4). There was no difference normotensive vessels the decrease was earlier (P among capillaries attributable to layer (OPL vs. IPL). < 0.001 between early and intermediate). PCP expresses the percentage of endothelial circum- AVP, the total content of filamentous actin within ference covered by directly adjacent (primary) peri- all areas of viable pericyte cytoplasm, was obtained by cyte processes. It is a proportion characteristic of two measuring the physically smallest parameter, that is, length measurements (the sum of PLP along IBML). the sum of individual AA, and comparing it to the In normotensives, PCP was 41% ± 10%, in hyperten- overall VPA, another physically small parameter (Fig. sives 48% ± 10%. 5). Because of their small size, both parameters are Similarly, actin coverage (ACP), the sum of ALP subject to a larger proportionate level of measure- along the IBML within primary pericyte processes, ment noise, leading one to expect some degree of in- was increased in SHR capillaries over WKY vessels stability. The AVP pattern of hypertensive capillaries (Fig. 4), again with no layer-specific change. The dif- was at a consistent distance above the pattern of nor- ference between hypertensive and normotensive ves- motensive capillaries, but quantitative significance sels was most significant at 44 wk (P < 0.001). Values at could not be effectively established (P values for the 24 wk were P = 0.002, and at 68 wk P = 0.008. ACP in three age groups were 0.15, 0.003, and 0.45, respec- normotensives was 26% ± 9%, and in hypertensives tively). Layer or age effects were absent. 32% ± 9%. ACA, the combined actin filament area in all peri- The total area of viable pericyte cytoplasm (pri- cytes relative to the WA, showed significance only at mary and more external processes counted together) the intermediate age group (P < 0.001). Significant relative to the overall wall area (VPCA) was signifi- levels between the early and late groups of hyperten- cantly larger in hypertensive than in normal vessels (P sive and normotensive vessels were P = 0.01 and 0.30, < 0.001) only in the intermediate age group of 44 wk respectively (Fig. 5). In ACA, the denominator WA is a

Downloaded from iovs.arvojournals.org on 09/27/2021 426 Investigative Ophthalmology & Visual Science, February 1993, Vol. 34, No. 2

PCP (PL/IBML) ACP (ALP/IBML)

O in CO d

to o CO d

CO CM o d

CM

20 30 40 50 60 70 20 30 40 50 60 70 Age in Heeks Age in Neeks FIGURE 4. Pericyte coverage (PCP) and actin coverage (ACP) were increased in SHR (1) over WKY (0).

physically larger parameter with proportionately less ± 0.02% (SEM).10 Third, Sosula and coworkers mea- measurement noise than VPA in the previously men- sured retinal capillaries of rats at 100 days of age, but tioned AVP. Also, AA constitutes a larger proportion using immersion fixation.11 Their total capillary area in AVP than in ACA. This may in part account for the was between 17.5 ±6.7 /xm2 (IPL) and 20.9 ±8.7 ^m clearer result regarding the latter. (OPL), compared to our OBMA of 18.5 nm (pooled data, Table 2). Their total pericyte cell area of 3.4 ±1.6 Mm2 (IPL) and 3.6 ± 3.2 ^tiii2 (OPL) compares to DISCUSSION our value for viable pericyte cytoplasm area (VPA) of Our technique of processing and measuring retinal 2.3 fxm2. Data for PCP are not included in their report. capillaries is significantly different from the tech- This survey of available comparative data indicates niques of others who have measured rat retinal capil- that our proportion characteristic PCP is nearly iden- laries in detail. This technical difference could influ- tical to that obtained by others with better techniques ence our data. Actin decoration with myosin subfrag- of tissue preservation. Also, our presumably artifact- ment-1 necessarily induces tissue artifacts that disrupt induced difference in absolute measurements stayed and obscure cell borders, yet preserve endothelial and within limits sufficient to interpret the comparative pericyte basement membranes. Other investigators effects of systemic hypertension on the retinal micro- have used perfusion fixation, which currently provides vasculature, provided that hypertension itself did not the best possible preservation of ultrastructural detail, induce significant new factors that may have in- although these authors have not provided actin data. fluenced measurement techniques. According to the Tilton et al measured in normal adult rats (weight data reported herein, sustained systemic hypertension 350 g) the "capillary circumference," a parameter affects the capillary bed of the retina uniformly, re- closely corresponding to our OBML.9 Their value was gardless of location closer to the arterial inflow (capil- 18 Mm. ours 16 nm (pooled data, Table 2). His pericyte laries of the more superficial IPL) or closer to the coverage (PCP) was 47.5% ± 6.8%, ours 41% ± 10%. venous outflow (capillaries of the more deeply located Frank et al, also working with perfusion fixation and OPL). The most significant information comes not rats older than 1 year, found a PCP value of 41% from pooled results (Table 2), but from longitudinal

Downloaded from iovs.arvojournals.org on 09/27/2021 Retinal Pericytes in Hypertension 427

VPCA (VPA/NA) AVP (AA/VPA)

CO - o /

CM / CO - O

O CO - d ).28

20 30 40 50 60 70 20 30 40 50 60 70 Age in Weeks Age in Neeks

ACA (AA/WA) 0.18

CD d 0.14

CM d

20 30 40 50 60 70 Age in Weeks

FIGURE. 5. At the intermediate time point, total viable pericyte area relative to total wall area (VPCA) was increased in SHR (1). The total actin in viable SHR pericytes (AVP) seemed increased over WKY (0), as was the total actin relative to the wall area (ACA).

observations comparing hypertensive and normoten- WT at all time points and an early increase in WA sive vessels at three time points, early (24-wk-old ani- (apparent increase at intermediate time point, P mals), intermediate (44 wk), and late (68 wk). = 0.002, no increase at late time point). Furthermore, After an apparent early dilation (P < 0.002), hy- PCP is significantly larger at early and intermediate pertensive capillaries show significant constriction at stages, and the total area of viable pericyte cytoplasm the intermediate stage compared to normotensive cap- relative to the vessel wall area (VPCA) is significantly illaries (P < 0.001). In the late stage, hypertensive ves- increased at the intermediate stage. sels remain constricted, whereas normotensive vessels All changes mentioned could be interpreted as show an age-dependent significant dilation. Hyperten- signs of pericyte hypertrophy or of pericyte hyperpla- sion also produces significant increases in capillary sia. For , one would need evidence that the

Downloaded from iovs.arvojournals.org on 09/27/2021 428 Investigative Ophthalmology & Visual Science, February 1993, Vol. 34, No. 2

number of pericytes is increased, for example as pericytes contain myosin and tropomyosin.19"21 All shown by an increased number of pericyte nuclei these cellular specializations of pericytes are similar to counted in a trypsin digest preparation. This was done those of smooth muscle cells and have been inter- in a previous study in monkey where the stress of reti- preted as evidence for a pericyte-mediated contractil- nal branch vein occlusion on capillary collaterals was ity of the microvascular wall. In that context, our find- evaluated.5 In that study, the pericyte number re- ing of hypertension-associated pericyte changes with mained stable, and pericyte changes were interpreted actin increase would mean that the pericyte-mediated as hypertrophy. In this study of systemic hypertension contractile potential is increased and responsible for in rats, no pericyte nuclei were counted. Even if a the observed capillary constriction. count had been attempted, results likely would have Other in vivo evidence for pericyte contractility is been equivocal because pericyte nuclei in rat retinal sparse, although the study by Tilton et al is very sug- digest preparations are less distinct than in monkey gestive.22 They perfused capillaries with the vasocon- and cannot be as clearly distinguished from endothe- strictive agents angiotensin, norepinephrine, and va- lial cell nuclei. In the resting retinal capillaries of mice, sopressin, and observed perfusion pressure increases pericytes constitute a very stable cell population with- that were higher in capillaries with tight pericyte cover- 12 out evidence for mitotic division, and in retinal capil- age compared to capillaries with less dense coverage. laries of SHR rats there was reportedly no difference Pressure increase was assumed to result from higher in form and number of pericytes compared to WKY vasoconstrictive vascular resistance caused by peri- 13 capillaries. The latter report, however, did not pro- cytes, but the vasoactive substances also could have vide quantitative data. acted directly on endothelial cells, causing them to Extensive work by another group found very good narrow the vascular lumen, or on smooth muscle cells evidence that the SHR condition is associated with pro- in larger precapillary vessels. liferation of pericytes, at least in the . In tissue Development of and of actin in- culture of hypertensive rat brain, microvascular peri- crease apparently is slow to involve the capillary bed, cytes were identified by immunostaining with actin- because our most significant data were obtained at the specific antibodies. As hypertension developed in the intermediate stage. Early hypertensive retinal capillar- rats, the ratio of pericytes to endothelial cells steadily ies seemed to dilate, possibly because hypertensive increased two to five times.14 Furthermore, in situ ex- stress was already transmitted, yet actin increase may periments showed that immunostaining was largely lo- have been insufficient to affect capillary caliber. calized to microvascular pericytes, and that capillaries Constriction of the microvasculature is consistent of certain brain regions (brain stem, motor cortex) with the general nature of SHR hypertension. This were more richly endowed with pericytes in SHR than condition is considered by many to be the model clos- in WKY rats.15 These studies suggest that microvascu- est to essential hypertension in humans.23 Hyperten- lar pericytes of the can prolif- sion activates vascular protein synthesis, which then erate, and may be involved in the development of mi- becomes the common metabolic pathway to structural crovascular disease. Because the authors also found a change.24 For example, Gabbiani et al observed endo- topographic variation of pericyte density among dif- thelial hypertrophy and hyperplasia, and an increase ferent brain regions of both SHR and WKY animals,15 in cytoplasmic actin filaments, suggesting that in re- pericytes also may play an important role in regulating sponse to hypertension the vascular may normal microvascular blood flow. Considering this lit- adaptively produce a cytoplasmic contractile appara- erature, we assume that the pericyte changes of our tus similar to that present in smooth muscle.25 Arterial current experiment may reflect hyperplasia. vessels develop a thicker media than comparable WKY Hypertensive pericyte changes include pericyte vessels, due to hypertrophy and hyperplasia of smooth actin content. Both ACP and ACA in the pooled data muscle cells. The result is greater peripheral microvas- (Table 2) are significantly increased, most clearly at the cular resistance and a heightened vasoconstrictor re- intermediate stage (Figs. 4, 5). Hypertensive pericyte sponse.26 changes associated with an increased contractile po- On the other hand, it was initially somewhat sur- tential of the vessel wall could explain the nature and prising to find that hypertension caused early dilation time course of capillary constriction. and later constriction of retinal capillaries 4 to 5 /mi in In the normal retina, capillary pericytes are diameter. In the central nervous system, the arterial uniquely frequent and contract in vitro during normal blood vessels through their autoregulation are ex- growth16 or in response to magnesium adenosine tri- pected to protect the microvasculature. The smallest phosphate.17 They also relax in response to the actin blood vessels reported to show hypertensive changes filament-disrupting agent, cytochalasin B.16 Pericytes on histometric observation were human cerebral arter- are linked to endothelial cells by adhesion plaques18 ies with an internal diameter of 10 /mi.27 These vessels and by direct cytoplasmic junctions.2 Biochemically, had measurable thickening of the media. The study,

Downloaded from iovs.arvojournals.org on 09/27/2021 Retinal Pericytes in Hypertension 429

however, used human autopsy tissue where hyperten- would have to be activated indirectly, through interac- sion had been longstanding and severe (diastolic pres- tion with the endothelium. sures consistently exceeded 110 mmHg). On the other hand, binding sites localized to cells Arteriolar protection of the capillary bed may be other than the endothelium may not be accessible to reduced under less severely hypertensive conditions. blood-borne mediators because of the tight blood-ret- Rats with short-range surgical28 or chemical hyperten- inal barrier. In that case, local synthesis or activation sion7 (systolic blood pressure >200 mmHg lasting 2 to of mediators may be advantageous and be accom- 9 wk) showed increased retinal capillary permeability plished by an ocular renin-angiotensin system. All of to peroxidase28 or to lanthanum.7 There also were such a system's recognized components have been patchy smooth muscle necroses in small retinal arteri- identified in the eye, and at least two of them, prore- oles and occasional degenerative pericyte changes. A nin30 and angiotensinogen,31 are prominently present similar observation was made in cynomolgus monkeys in the ciliary body. It would be interesting to see if that had systolic blood pressures of > 150 mmHg plus research with higher-resolution methods can localize average pressure rises of >43 mmHg, lasting 1 week to binding sites for vasoactive mediators to the external 4 months. The authors saw degenerative changes of layer of the retinal capillary wall (ie, to the enigmatic retinal capillaries involving pericytes alone, or peri- pericyte). cytes and endothelial cells. Rare increases of pericyte volume also were suspected.6 Although these results Key Words on rats and monkeys were not based on systematic mea- actin filaments, morphometry, pericytes, spontaneously hy- surements, they nevertheless suggest that early hyper- pertensive rats, systemic hypertension. tension increases intraluminal pressure and impairs ar- teriolar contractility. This may expose the retinal capil- A cknowledgmen ts lary bed to abnormal pressure, resulting in both The authors thank Marion Greaser (Muscle Biology Labora- degenerative and adaptive hypertrophic-hyperplastic tory) for providing substantial quantities of freshly prepared changes. Hence, documentation of the latter in our myosin subfragment-1, Grelchen Poulsen and Cara Blem- detailed quantitative evaluations of SHR rats is cer- ings for special expertise in monitoring the blood pressure tainly within the expected pattern of hypertensive vas- values of the experimental animals, and Cara Blemings and cular disease. Anna Vallo for preparing photographic montages and per- forming multitudes of planimetric measurements. One would like to speculate that in the retina, hy- pertensive pericyte changes with actin increase and References capillary constriction represent an exaggerated re- 1. Wallow IHL, Burnside B. Actin filaments in retinal sponse of normal pericyte function. In this sense, our pericytes and endothelial cells. Invest. Ophthalmol Vis observations provide further in vivo indirect support Sci. 1980; 19:1433-1441. for the theory that pericytes, through their contractil- 2. Matsusaka T. Ultrastructural differences between the ity, may influence microvascular caliber and blood choriocapillaries and retinal capillaries in the human (low. For such contractile function to be meaningful, a eye. Nippon Ganka Gakkai Zasshi. 1969; 73:1603- signaling system through nerves or blood-borne media- 1605. tors is needed. 3. Hudes GR, Li W, RockeyJH, White P. Prostacyclin is the major prostaglandin synthesized by bovine retinal Because autonomic nerve endings are notably ab- capillary pericytes in culture. Invest Ophthalmol Vis Sci. sent from the retina, blood-borne mediators may play 1988;29:1511-1516. an important role. Observations of others become in- 4. Ferrari-Dileo G, Davis EB, Anderson DR. Effects of triguing (ie, that cultured retinal pericytes seem to pos- cholinergic and adrenergic agonists on adenylate cy- 4 sess functional adrenergic and cholinergic receptors ). clase activity of retinal microvascular pericytes in cul- Angiotensin II binding sites also were identified.29 ture. Invest Ophthalmol Vis Sci. 1992;33:42-47. These were seen mostly in the capillaries of the optic 5. Wallow IHL, Bindley CD, Linton KLP, Rastegar D. nerve head, and retinal capillaries showed little, if any, Pericyte changes in branch retinal vein occlusion. In- evidence for the presence of such receptors. The au- vest Ophthalmol Vis Sci. 1991; 32:1455-1463. thors used autoradiography combined with light mi- 6. Garner A, Ashton N, Tripathi R, Kohner EM, Bulpitt croscopy, however, a method that has limited resolu- CJ, Dollery CT. Pathogenesis of hypertensive retinopa- thy: an experimental study in the monkey. BrJ Ophthal- tion and may fail to detect binding sites accurately in mol. 1975; 59:3-44. small retinal capillaries. This method also was insuffi- 7. Forthomme D, Cantin M. The retinal capillaries of the cient to allow specific localization of angiotensin II rat in deoxycorticosterone hypertension. Am J Pathol. binding sites within the layers of the microvascular 1976;85:263-276. wall. If binding sites were localized to the vascular en- 8. Wallow IHL, Greaser ML, Stevens TS. Actin filaments dothelium, various blood-borne mediators would have in diabetic fibrovascular preretinal membrane. Arch direct access, and smooth muscle cells and pericytes Ophthalmol. 1981; 99:2175-2181.

Downloaded from iovs.arvojournals.org on 09/27/2021 430 Investigative Ophthalmology & Visual Science, February 1993, Vol. 34, No. 2

9. Tilton RC, Miller EJ, Kilo CH, Williamson JR. Peri- in pericytes: II. immunocytochemical evidence for the cyte form and distribution in rat retinal and uveal cap- presence of two isomyosins in graded concentrations. illaries. Invest Ophthalmol Vis Sci. 1985;26:68-73. J Cell Biol. 1985; 100:1387-1395. 10. Frank RN, Dutta S, Mancini MA. Pericyte coverage is 22. Tilton RG, Kilo C, Williamson JR, Murch DW. Differ- greater in the retinal than in the cerebral capillaries of ences in pericyte contractile function in rat cardiac the rat. Invest Ophthalmol Vis Sci. 1987; 28:1086- and microvasculatures. Microvasc Res. 1091. 1979; 18:336-352. 1 1. Sosula L, Beaumont P, Jonson KM, Hollows FC. 23. Friedman S. Vascular reactivity. In: Genest J, Kuchel Quantitative ultrastructure of capillaries in the rat ret- O, Hamet P, Cantin M, eds. Hypertension: Physiopathol- ina. Invest Ophthalmol Vis Sci. 1972; 11:916-925. ogy and Treatment. 2nd ed. New York: McGraw-Hill; 12. Engerman RL, Pfaffenbach D, Davis MD. Cell turn- 1983:457-473. over of capillaries. Lab Invest. 1967; 17:738-743. 24. Yamori Y. Physiopathology of the various strains of 13. Yoshimoto H, Matsuyama S. Structural characteristics spontaneously hypertensive rats. In: Genest J, Kuchel of the pericyte in the retinal capillary. Japanese Soci- O, Hamet P, Cantin M, eds. Hypertension Physiopathol- ety for Abstracts. Microvasc Res. ogy and Treatment. 2nd ed. New York: McGraw-Hill; 1982;24:230. 1983:556-581. 14. Herman IM, Nevvcomb PM, Coughlin JE, Jacobson S. 25. Gabbiani G, Badonnel MC, Rona G. Cytoplasmic con- Characterization of microvascular cell cultures from tractile apparatus in aortic endothelial cells of hyper- normotensive and hypertensive rat : pericyte- tensive rats. Lab Invest. 1975;32:227-234. endothelial cell interactions in vitro. Tissue Cell. 26. Roy JW, Mayrovitz HN. Microvascular pressure, flow, 1987; 19:197-206. and resistance in spontaneously hypertensive rats. Hy- 15. Herman IM, Jacobson S. In situ analysis of microvas- pertension. 1984; 6:877-886. cular pericytes in hypertensive rat brains. Tissue Cell. 27. Cook TA, Yates PO. A histometric study of cerebral ]988;20:l-12. and renal in normotensives and chronic hy- 16. Kelley C, D'Amore P, Hechtmann HB, Shepro D. Mi- pertensives. J Pathol. 1972; 108:129-135. crovascular pericyte contractility in vitro: comparison with other cells of the vascular wall. J Cell Biol. 28. Giacomelli F, Juechter KB, Wiener J. The cellular pa- 1987; 104:483-490. thology of experimental hypertension: VI. alterations 17. Das A, Frank RN, Weber ML, Kennedy A, Reidy C, in retinal vasculature. Am]Pathol. 1972;68:81-96. Mancini MA. ATP causes retinal pericytes to contract 29. Ferrari-Dileo G, Davis EB, Anderson D. Angiotensin in vitro. Exp Eye Res. 1988;46:349-362. II binding receptors in retinal and optic nerve head 18. Courtoy PJ, Boyles J. in the microvascula- blood vessels. Invest Ophthalmol Vis Sci. 1991; 32:21— ture: localization in the pericyte-endothelial intersti- 26. tium.y Ullrastruct Res. 1983;83:258-273. 30. Sramek SJ, Wallow IHL, Day RP, Ehrlich EN. Ocular 19. Gordon SR, Essner A. Actin, myosin, and laminin lo- renin-angiotensin: immunohistochemical evidence for calization in retinal vessels of the rat. Cell Tissue Res. the presence of prorenin in eye tissue. Invest Ophthal- 1986; 244:583-589. mol Vis Sci. 1988;29:1749-1752. 20. Joyce NC, Haire MF, Palade GE. Contractile proteins 31. Sramek SJ, Wallow IHL, Tewksbury DA, Brandt CR, in pericytes: I. immunoperoxidase localization of Poulsen GL. An ocular renin angiotensin system: im- tropomyosin.y Cell Biol. 1985; 100:1379-1386. munohistochemistry of angiotensinogen. Invest Oph- 21. Joyce NC, Haire MF, Palade GE. Contractile proteins thalmol Vis Sci. 1992;33:1627-1632.

Downloaded from iovs.arvojournals.org on 09/27/2021