increase in endothelial cell density before artery ... · pdf fileshunt was closed after the...

Increase in Endothelial Cell Density before ArteryEnlargement in Flow-loaded Canine Carotid Artery

Hirotake Masuda, Koichi Kawamura, Kohei Tohda, Takesi Shozawa,Masato Sageshima, and Akira Kamiya

To Investigate the morphologic basis of blood flow-dependent adaptive vascularenlargement, we quantltated endothelial cell density, dimensions, and structure incanine carotid arteries that were flow-loaded for 4 weeks, i.e., Just before thedevelopment of significant adaptive enlargement Increased flow was produced Inthe right common carotid artery of seven adult beagle dogs by an arterlovenousshunt to the right external Jugular vein. The left common carotid artery was used toproduce sham-operated controls. Five additional animals were used to producesham-shunted controls, and two dogs were used as nonoperated controls. The bloodflow rate (BFR) and wall shear rate (WSR) were markedly Increased immediately afteranastomosis In the proximal segment of the shunted artery (BFR=719±142 ml/min,WSR>4127±1002/sec) and after 4 weeks (BFR=628±157 ml/mln, WSR>2919±388/sec) compared to the same artery before anastomosis (BFR=154±50 ml/mln,WSR=904±314/sec, p<0.01x10~3 for both comparisons) and to the contralateralcontrol artery after 4 weeks (BFR=365±110 ml/mln, WSR=2136±876/sec, p<0.01and p<0.05, respectively, compared to the shunted side). In the shunted artery,endothelial cell density was markedly Increased (6.15±0.68x103 cells/mm2 comparedto 3.33±0.70x103 cells/mm2 for the controls, p<0.001). Endothelial cells on the highflow side were markedly narrowed In both axial and circumferential directions, butwere radially thickened; nuclei became prolate-spheroid in shape. On the controlside, cells were relatively flat and thin. We conclude that elevated wall shear stressInduces an early Increase in endothellal cell number and that this Increase precedesthe development of significant blood flow-dependent vascular enlargement(Arteriosclerosis 9:812-823, November/December 1989)

A continuous lining of intact endothelial cells is main-tained in arteries under conditions of normal flow.1-6

It has become increasingly evident that endothelial cellsundergo configurational modifications in response to flowin vitro789 and in vivo.101112 Both the shape and orien-tation of endothelial cells have been related to localhemodynamic conditions. Blood flow-dependent adaptivevascular growth and dilation, first pointed out by Thoma in1893,13 has recently been demonstrated to be closelyassociated with wall shear stress in experimental animalmodels.1415 Although the response is thought to bemediated by endothelial cell reactions to shear, studieshave thus far dealt with the flow-related changes in arteryradius,14-22 which appear to stabilize when wall shearstress values are equal to about 15 dynes/cm2.14'15 Tofurther characterize the participation of endothelial cells inthe process, we studied endothelial cell density andstructure in canine carotid arteries subjected to anincreased flow load by an arteriovenous (AV) shunt. The

From the Second Department of Pathology, AWta UniversitySchool of Medicine, AWta, and the Research Institute of AppliedElectricity, Hokkaido University, Sapporo Japan.

This study was supported by Research Grant 60-C2 for Car-diovascular Diseases from trie Ministry of Health and Welfare andResearch Grant 61570164 from the Ministry of Education, Japan.

Address for correspondence: Hirotake Masuda, M.D., SecondDepartment of Pathology, Akita University School of Medicine,1-1-1 Hondo, Akita 010, Japan.

Received January 25, 1988; revision accepted June 12, 1989.

present report deals with the changes evident at 4 weeksafter anastomosis, just before significant shear stress-dependent adaptive dilation became evident.14

MethodsAnimals

Fourteen female beagle dogs, 12 to 18 months old, eachweighing 7.0 to 12.5 kg, were used for these experiments.Anastomosis between the right carotid artery and the rightexternal jugular vein was performed on seven animals(ages, 15.6±2.4 months; weights, 10.7±1.6 kg). All theoperative procedures except for shunting were performedon the left side in the same animals to control for operativemanipulation without anastomosis (sham-operated con-trols). To examine the effects of the complete shuntingprocedure, five dogs were used as a second set of controls.In these animals, the anastomosis was performed, but theshunt was closed after the operation and before the hemo-static clamps were released. These animals were kept for1 week (three animals) or for 4 weeks (two animals). Twodogs served as nonoperated controls (ages, 15 and18 months; weights, 9.5 and 12 kg). The same hemody-namic measurements, fixation techniques, sampling, andmorphologic studies were performed on all the animals.

Although canine right and left carotid arteries are ofdifferent lengths (the left branches from the brachioce-phalic artery at about 2 cm from the aortic orifice, while the

812

by guest on April 2, 2017

http://atvb.ahajournals.org/D

ownloaded from

http://atvb.ahajournals.org/

ENDOTHELIAL RESPONSE TO INCREASED FLOW Masuda et al. 813

right branches at about 1 cm from the left), we elected toperform all of the anastomoses on the right side and touse the vessels on the left as the sham-operated controls.Preliminary studies of 20 adult female beagle dogs in ourlaboratory revealed that the outer diameter was the samefor the right and left carotid artery (3.5 mm) and that bloodflow was identical for both sides: 146±34 ml/min on theright and 147±29 ml/min on the left (r=0.97, p<0.001).The flow patterns in the two vessels were also indistin-guishable. In addition, detailed ultrastructural observa-tions on four adult female beagle dogs showed no differ-ences between the right and left arteries.

Operative ProceduresThe seven dogs subjected to shunting were anesthe-

tized with sodium pentobarbital (30 mg/kg). The rightcommon carotid artery and the right external jugular veinwere exposed, and an AV shunt was constructed understerile conditions as described by Kamiya and Togawa.14

The AV anastomoses were performed at a distance of10 cm from the origin of the artery. On the left, thecommon carotid artery was sutured to the left externaljugular vein along the adventitia after having been sub-jected to the same manipulation and clamping as on theright. These vessels served as controls for operativemanipulations without anastomosis (sham-operated con-trols). Measurements of blood flow rate (BFR) were per-formed before operation and immediately after operationproximal to the anastomotic site on the shunted side andat the same level on the control side by using an electro-magnetic flowmeter (Nihon Kohden Company, MFB 2100,Japan). These animals were kept for 4 weeks, and thethrill in the region of the AV anastomosis was checkeddaily. For the final measurements at 4 weeks, the animalswere anesthetized as before, the vessels were exposed,and measurements were performed on both sides at thesame levels as previously. The aortic pressure was deter-mined by a catheter introduced into the thoracic aorta byway of the femoral artery and connected to a strain-gaugemanometer (Statham, P231D, Gould Inc., Cleveland, OH).

Tissue PreparationFor light microscopic and ultrastructural studies, the

common carotid arteries were fixed in situ while dis-tended. In six of the shunted and in the two nonoperatedanimals, controlled pressure perfusion-fixatjon was car-ried out as described previously20 by using 3% glutaral-dehyde in phosphate buffer (pH 7.4). The height of aninflow reservoir and the outflow resistance were adjustedto maintain a constant perfusion pressure of 100 mm Hg.After 30 minutes, a segment of the common carotid artery,10 mm in length, was resected just proximal to the site ofthe blood flow measurement on the shunted side at 7 cmdistal to the origin of the artery and at the correspondinglevel on the control side (i.e., 8 cm distal to the origin).One-half of the length of the resected specimen was usedfor light microscopy. Precise transverse sections wereembedded in paraffin, sectioned at 1.5 /un, and stainedwith hematoxytin and eosin.

For scanning electron microscopy (SEM), one-fourth ofthe resected segment, 2.5 mm in length, was postfixed for

1 hour in 3% glutaraldehyde in phosphate buffer (pH 7.4)at 4°C. This was dehydrated through alcohols and wascritical-point dried. The samples were coated by evapo-rating gold-platinum in a high vacuum. The endothelialsurface was observed in a JSM-T200 (JEOL Company,Japan) scanning electron microscope.

For transmission electron microscopy (TEM), the remain-ing one-fourth of the resected segment was postfixed for1 hour in 3% glutaraldehyde in phosphate buffer (pH 7.4)at 4°C and then for 1 hour in 1% osmic acid in phosphatebuffer (pH 7.4) at 4°C. Specimens were transected in bothaxial and transverse directions to obtain precisely orientedlongitudinal and transverse planes of section. After dehy-dration through alcohols and embedding in Epon, ultrathinsections were stained with lead citrate and uranyl acetateand were observed in an LEM 2000 (Akashi Company,Japan) transmission electron microscope. Samples fromthe seventh dog with an AV fistula were used to obtainsilver deposit preparations to mark intercellular junctions.For this preparation, the carotid arteries were flushed in situwith 5% dextrose, were infused with 20 ml of 0.25% silvernitrate in 5% dextrose for 30 seconds, were flushed withthe same volume of 5% dextrose, and then were perfusedwith normal saline solution. Fixation under pressure wassubsequently performed, and the tissues were sampled forlight microscopic and SEM in the manner described above.

Hemodynamlc ComputationsTo calculate the hemodynamic parameters, we obtained

the internal radius (r) using the expression:

r=1.25L/27r, 0)where L is the luminal circumference on transverse sec-tion and 1.25 is a correction factor for shrinkage occurringduring fixation, dehydration, and paraffin embedding.23

The circumference was determined by projecting thetransverse section of the artery onto a sheet of paper at anenlargement of 100x (Nikon Company, V-16A, Japan),then tracing the lumen contour and measuring the circum-ference by means of an image analyzer system (NikonCompany, Cosmozone-1, Japan). For purposes of com-putation, the internal radii of the shunted (right) arteriesand of the control (left) arteries before operation andimmediately after operation were assumed to be the sameas those of the control (left) arteries at 4 weeks afteroperation.

Density of Endothelial Cell PopulationThree complimentary methods were used. We deter-

mined the mean of the total number of endothelial cellnuclei visible along the entire circumference on 10 suc-cessive transverse light microscopic sections. In addition,the number of endothelial cells and nuclei per unit ofcircumferential length was obtained from counts on acomposite of 20 to 30 TEM photographs encompassing a0.4 mm segment of the circumference of a transversesection. The photographs were taken at a magnification of10 OOOx. The abluminal width of each endothelial cell:

wC,, wC2 wCn.,, wCn (2)



ownloaded from


814 ARTERIOSCLEROSIS V O L 9, No 6, NOVEMBER/DECEMBER 1989

was measured, and the total abluminal length of thecomposite segment (WC) was then obtained:

WC=wCl+wC2+...+wCn.1+wCn (3)

where n=the total number of endothelial cells in onecomposite transverse TEM photograph. The number ofendothelial cells per millimeter of length (1000xn/WC)was then calculated. Similarly, trie number of endothelialnuclei per millimeter of length (1000 x m/WC, where m=thetotal number of endothelial nuclei in one composite,transverse TEM photograph) was determined from theultrathin transverse sections. In each composite, trans-verse TEM photograph, 100 to 150 endothelial cells and15 to 20 endothelial nuclei in the control common carotidartery and 150 to 200 endothelial cells and 20 to 30 en-dothelial nuclei in the shunted common carotid arterywere measured. Finally, the number of endothelial cellsper unit of lumen surface area was assessed from SEMphotographs taken at a magnification of 1000x. Thenumber of endothelial cells on a photograph divided bythe area of the photograph was determined for eachphotograph. The cell density (cells/mm2) for each arterywas expressed as the average of the determinations from10 SEM photographs for each artery.

Size and Shape of Endothelial CellsSeveral assumptions were made in obtaining estimates

of the shape and size of endothelial cells and nuclei: 1) Allthe endothelial cells and nuclei in a given common carotidartery specimen were the same shape and size. 2) Thelong axes of the cells were oriented parallel to thelongitudinal axes of the blood vessels. 3) Cell body andnuclear shapes were symmetrical in both transverse andlongitudinal dimensions. 4) The distribution of these struc-tures was random. When the endothelial layer is preciselytransected, transversely or longitudinally, each nuclearand cellular profile represents part of a nucleus and a cellat a given level. TEM ultrathin sections are very thin(approximately 70 nm) compared to the size of an endo-thelial cell nucleus (10 to 15 /un in length). Therefore,endothelial cell and nuclear profiles appearing on a com-posite, transverse TEM photograph enlarged 10 000xwere used to estimate the dimensions of a single endo-thelial cell and nucleus within a constant thickness ofsection d (^m), where d=LE/n, and LE=length of anendothelial cell (fim).2* For the same composite photo-graph, d is the same for both cell and nucleus:

d=LE/n=LN/m (4)

where LN=length of endothelial nucleus (nm). The sameconsiderations are applicable to the longitudinal sections.In practice, however, it was difficult to identify endothelialcell junctions and, due to the elongated spindle shape ofthe endothelial cells, to obtain a sufficient number of cellsand nuclei in a composite longitudinal TEM photographcovering 0.4 mm of actual length. Therefore, we used thelongitudinal sections to determine the length of endothelialnuclei (LN in microns) by measuring the longest nuclearprofile on a composite longitudinal TEM photograph. Fromthese data, we obtained the d value for d=LN/m.

To measure endothelial cell and nuclear profile areason transverse TEM photographs, we resorted to pointcounting with 5x5 mm grids.21 The endothelial profileareas of successive endothelial cells in the composite,transverse TEM photographs were taken as: aC,, aC2

aCn., and aCn (fim2) and the successive nuclear profileareas as: aN,, aN2 aNm.,, and aNm (pm2). Reconstruc-tions of endothelial cell and nuclear profiles were obtainedby superimposing all of the transverse profiles appearingin a composite, transverse TEM photograph according tothe incremental size of endothelial cells and nuclearwidths within a distance of 0.5xd along a lengthwisecentral index line. The endothelial abluminal area, CA(jtm2), was obtained from:

d (wC1+wC2+...+wCn.,+wCn)=dxWC=LNxWC/m (5)

The endothelial volume CV (/xm3) was obtained from:

d (aC1+aC2+...+aCn.1+aCn)=LNxAC/m (6)

where AC was the sum of endothelial cell section profileareas (AC=aC,+aC2+...+aCn.1+aCn) appearing in a trans-verse composite TEM photograph. The endothelial nucleararea (NA, ysn2) was obtained from:

dxWN = LNxWN/m, (7)

where WN was the sum of widths of endothelial nucleiappearing in a composite transverse TEM photograph(AN=aN1+aN2+...+aNnM+aNm).

Although these assumptions were justified by the find-ings in canine carotid arteries given in a previous report19

and by the SEM findings in the present experiments,endothelial cells actually showed some variation in angleof orientation, size, and shape. According to the quantita-tive data on endothelial orientation, size, and shapefurnished by Cornhill et al.1 in the casts of rabbit aortas,nearly 95% of endothelial cells were oriented in their longdimension at an inclination within 25° of the axial directionof the vessel, and 100% were oriented within 45°. In thesame study, 60% of endothelial cells were found to be600 to 900 nm2 in area, and 80% had a width/length ratioof 0.2 to 0.4. If we assume that the data of Cornhill et al.are similar for all large arteries and, therefore, for thecanine carotid artery as well, the area obtained from ourexpression, dxWC, is no more than 1.12 larger than theactual value, because 0.95/cos 25°+0.05/cos 45°=1.12,and dxWCx1.12 gives the maximum value for the ablu-minal area. Referring to the histogram (in steps of 100LUTI2) of endothelial surface areas provided by Cornhill etal., we note that we had more than 100 cut profiles in eachof our composite photographs and a sufficient number (7to 20) of sections to match each histogram step fromCornhill et al. Because the number of cells in each of thehistogram steps is expected to correspond to the fre-quency of occurrence of each of the cut section profiles,our equation for area, dxWC, will give the mean area ofendothelial cells. Similarly, the volume, length, and widthof cells obtained from our equations will be mean values.There are no comparable published data for nuclei, butthe nuclear deviation angles should be almost the sameas for cells. From the literature,34 however, nuclei appearto be uniform in size and shape, and the number we



ownloaded from



Table 1. Hemodynamic Changes in Canine Carotid Arteries after Arteriovenous Anastomosis

Parameters

Internal radius*(cm)

Flow rate(ml/min)

Row velocity(cm/sec)

Reynolds number

Shear rate(1/sec)

Shear stress(dynes/cm2)

Before anastomosis

Shunted(right)

(0.157)

154±50

34±12368±118

904±314

27±9

Control(left)

(0.157)

154±48

35±11371±118

921 ±381

28±11

Just after anastomosis

Shunted(right)

(0.157)

719±142t+

156±31tt|1695±312ttB

4127±1002t*H

124±3Ot*j|

Control(left)

(0.157)

187±59

42±14449+139

1096±432

33±13

4 weeks after anastomosis

Shunted(right)

0.165±0.013

628±157t*

120±19t*1403±289t+

2919±388t§

88±12t§

Control(left)

0.157±0.018

365±110ttOH1l

81 ±27 t t t * l873±256**§§i|||

2136±876tt§§11

64±26tt§§W

Nonoperatedcontrols

(right and left)

0.172±0.027

173±29

35±16

393±123

891 ±506

27±15

All values are means±standard deviation. The Reynolds number is calculated using 0.03 poise for viscosity and 1.054 g/cm3for specificgravity. Shear rate and shear stress are calculated assuming Poiseuille flow.

'The value for internal radius used for computations for arteries before and immediately after anastomosis is shown in parentheses.The value was assumed to be the same as the value measured for the controls at 4 weeks. tValues on the shunted side are significantlylarger than before anastomosis (p<0.01 xiO"3). Values on the shunted side are significantly larger than the control side (+p<0.001,§p<0.05). Values just after anastomosis are significantly larger than at 4 weeks after anastomosis Q|p<0.005, Hp<0.05). Values on thecontrol side at 4 weeks after anastomosis are significantly larger than just after anastomosis (**p<0.005, ttp<0.005) or beforeanastomosis (#p<0.001, §§p<0.01), or in non-operated controls (||||p<0.01, Hp<0.05).

counted (15 to 20) was, therefore, considered to besufficient to arrive at a meaningful figure. From thesemorphometric data and using computer-assisted histo-grams of the endothelial width profiles, we obtained ablock-like drawing of the presumed shape of endothelialcells in each artery. On the basis of the SEM and TEMfindings, we performed "by eye" a simple smoothingoperation of the endothelial shape. Endothelial cell sur-face shape was also directly determined from the SEMphotographs of 20 cells of the silver-stained preparations.

Statistical AnalysisOur results are expressed as the means±the standard

deviations of the mean. Statistical analysis was performedby Student's t test. Differences were considered signifi-cant if p values were less than 0.05.

ResultsBlood Flow

The thrill at the AV anastomoses in the shunted animalspersisted in daily examinations throughout the experimen-tal period. The mean blood pressure was 134±12 mm Hgin the seven shunted animals and 130 and 150 mm Hg inthe two nonoperated controls. BFR and mean blood flowvelocities (D) (D=BFR/60-7rr2), with the value for lumenradius (r) obtained on the control side at 4 weeks are givenin Table 1. Immediately after anastomosis, BFR and D inthe shunted arteries were more than four times greaterthan before anastomosis (719±142 ml/min compared to154±50forBFRand 156±31 cm/sec compared to 34± 12for D; p<0.01 x10~3 for both). At 4 weeks after anastomo-sis, these values were about three times higher than beforeanastomosis (628±157 ml/min compared to 154±50 forBFR and 120±19 cm/sec compared to 34±12 for D;p<0.01x10"3for both). In one shunted artery, BFR andOreached the extremely elevated values of 960 ml/min and

215 cm/sec, respectively. BFR and D in the shuntedarteries immediately after anastomosis were significantlyhigher than in the control (left) arteries (719±142 ml/mincompared to 187±59 for BFR, p<0.01, and 156±31cm/sec compared to 42± 14 for D, p<0.05). BFR andU inthe control arteries at 4 weeks were, however, less mark-edly, but significantly, elevated compared to values incontrols before operation or just after operation (365±110ml/min compared to 187±59 for BFR and 81 ±27 com-pared to 42±14 for O; p<0.005 for both).

Shear Rate and Shear StressWhen the shear rate is about 1000/sec, as is normal for

many large blood vessels, non-Newtonian behavior becomesinsignificant and the apparent viscosity of blood approachesa value asymptotically in the range of 0.03 to 0.04 poise.28-26

By using 0.03 poise for blood viscosity1 *&& and 1.054g/cm3 for blood specific gravity,27 Reynolds numbers (Re)were calculated on the basis of the expression:

Re=1.054x2rxD/0.03 (8)

Re was nearly identical in the right and left commoncarotid arteries before operation (368±118, right and371+118, left). These values are well below the criticalvalue of 2000 at which turbulent flow occurs in a smoothstraight tube. In these arteries, the entrance length for fulldevelopment of the steady laminar flow profile was calcu-lated to be 3.5 cm, using the expression:

0.03x2rxRe (9)

where r is the radius and 0.03 is an experimentallydetermined constant.26 This computation is valid for Revalues between 10 and 2500. The entrance length valueof 3.5 cm is much shorter than the location of thesegments that form the basis of this report (7 cm on theright and 8 cm on the left). Wall shear rate (WSR) and wall



ownloaded from


816 ARTERIOSCLEROSIS VOL 9, No 6, NOVEMBER/DECEMBER 1989

Figure 1. Transverse sections of canine carotid arteries.A. Shunted artery. B. Control artery. The endothelial layer (E) Isthickened. Nuclear profiles (arrowheads) protrude into the lumenof the shunted artery. Endothelial cells are thin and flat on thecontrol side. Hematoxylin and eosin stain. x500

shear stress (WSS) for right and left were apparentlyidentical to the levels calculated assuming Poiseuille flow:

WSR=4(BFR)/60irr3 (10)

and

WSS=0.03 (WSR) dynes/cm2 (11)

After anastomosis, Re was high in the shunted arteries,reaching 1695±312 just after anastomosis and 1403±289at 4 weeks. In one shunted artery, Re was nearly 2300immediately after anastomosis. The entrance lengths justafter anastomosis and at 4 weeks proved to be about16 cm and 14 cm, respectively. The actual values of WSRand WSS would be expected to be larger than valuescalculated assuming Poiseuille flow, because the flowprofiles were flatter than would be present with a fullyestablished Poiseuille flow. Turbulent flow was likely in theone artery with Re equal to 2300. In control arteries afteroperation, Re and entrance length were 449+139 and4.2 cm, respectively, just after operation and 873±256and 8.2 cm at 4 weeks. In these arteries, WSR and WSSappeared to be identical to the levels calculated, assum-ing Poiseuille flow. The values for WSR and WSS shownin Table 1 were calculated with an assumption of Poi-seuille flow. In the shunted arteries, WSR and WSS werehigh (4127±1002/sec and 124±30 dynes/cm2) immedi-ately after anastomosis and were still at a high level(2919±388/sec and 88± 12 dynes/cm2) at 4 weeks. In thecontrol arteries, WSR and WSS were not significantlychanged immediately after operation, but were moder-ately elevated at 4 weeks (2136±876/sec compared to1096±432 for WSR, 64±26 dynes/cm2 compared to33±13 for WSS; p<0.005 for each).

Morphologic FeaturesThe endothelial layer was relatively thick, with nuclei

protruding prominently from the lumen surface in the

B

Figure 2. Intercellular junctions (arrows) on transverse sec-tions of canine carotid arteries. A. Shunted artery. B. Controlartery. Intercellular junctions are much more numerous in theshunted artery than in the control vessel. Silver preparation andhematoxylin and eosin stain. x500

shunted arteries (Figure 1A), while the endothelium wasflat and thin on the control side (Figure 1B). The intercel-lular junctions were much more numerous in shuntedarteries (Figure 2A) than in controls (Figure 2B). Endothe-lial cells on transverse sections of shunted arteries weremuch narrower than on the control side. Examination byTEM revealed the cells to be radially thickened, closelypacked, and drcumferentially narrowed in the shuntedarteries (Figure 3A and Figure 4). The nuclear contourswere round and protruded markedly into the lumen on thetransverse sections, and the cell surfaces showed occa-sional microvilli. The endothelial cells and nuclei on thecontrol side were flat (Figure 3B and Figure 4). Theintercellular junctions of shunted arteries were long andstraight on the transverse section (Figure 5A), while thoseof control arteries were short and convoluted (Figure 5B).Viewed by SEM, endothelial cells in the shunted arteryappeared slender with rounded, protruding surfaces intheir midportions (Figure 6A). Microvilli were observed atthe tips of these protrusions. Control arteries had rela-tively flat surface contours and few microvilli (Figure 6B).In shunted arteries, endothelial cells occasionally showeda constricted or cleaved nucleus, and some cells werebinucleate (Figures 7A and 7B). Pairs of endothelial cellsof nearly identical shape were occasionally arrangedsymmetrically in relation to sharp crevices (Figures 7Cand 7D). In the nonoperated controls, the endothelial cellswere flat and did not differ from the controls in lightmicroscopic or ultrastructural features. In the right com-mon carotid artery of the control sham-shunted animals,the endothelial cells were flat and did not differ in lightmicroscopic or ultrastructural appearances from nonoper-



ownloaded from



Figure 3. Transmission electron microscopic vtews of the endotheliai cell layer of canine carotidarteries. A. Shunted artery. B. Control artery. Endotheliai cells are thickened, and nuclei protrudemarkedly on the shunted side, while the cells are thin and flat in the control artery. i=intercellularjunction, n=nucleus. x3000

TRAHSVEBSE SECTIO* OF SHUKTn) AXT1BY

-»JT1 I—» H

10 alorosoter

I * HTBAJOVlEat 3ECTI0P OF COVTBOL AJTTIRY

Figure 4. Diagrammatic representation of the endotheliai layer of canine carotid arteries ontransverse section. wC=abluminal endotheliai cell width, aC=endothelial cross-sectional area,wN=endothelial nuclear width, aN=endothelial nucleus cross-sectional area.

ated controls. No differences were noted either at 1 weekor 4 weeks after operation.

MorphometryThe histometric data on endotheliai cell density are

given in Table 2. All five quantitative determinationsrevealed significant increases in the shunted arteriescompared to the contralateral control arteries and to thenonoperated controls. The total number of endotheliai cellnuclei on the transverse light microscopic sections of theshunted arteries was about 1.5 times greater than that onthe control side (488±69 nuclei compared to 317±40,p<0.001). Similarly, endotheliai cell density of shuntedarteries determined from the light microscopic and TEMmaterial was 1.5 times higher than in the controls (p<0.001and 0.0001, respectively). On SEM, the endotheliai celldensity of shunted arteries was 6.15+0.68X103 cells/mm2, which was 1.84 times greater than that of controlarteries (3.33±0.70x103 cells/mm2, p<0.01 x10~3).

Endotheliai cells lining the shunted arteries were nar-row and thick, in marked contrast to the wide and thinconfiguration in the controls. The dimensions of the endo-theliai cells are presented in Table 3. The endotheliai cell

density (4.1 X103 cells/mm2) as computed from the ablu-minal endotheliai cell area in the shunted arteries (241 ±45jtm2) was slightly lower than that obtained from SEMdeterminations (Table 2). This discrepancy is likely to bedue to differences in tissue shrinkage during processing,because the ratios of endotheliai cell density in shuntedarteries to that of controls obtained by the two differentmethods were almost identical. The endotheliai cell vol-ume in shunted arteries was 0.77 times that of the controls(p<0.05), while the volume relative to nonoperated con-trols was somewhat smaller. The average thickness ofendotheliai cells was 1.37 nm in shunted arteries, 1.00 jtmin the controls, and 0.82 ^m in the nonoperated controls.Although the average nuclear volume was the same forshunted and control vessels, the shape of endotheliai cellnuclei in shunted arteries was prolate-spheroid and that incontrol arteries, oblate-spheroid (Figure 8). The averagenuclear volume of the nonoperated control vessels wassomewhat smaller than for operated animals. The aver-age dimensions of endotheliai cells on the silver-stainedpreparations as studied by SEM were 49±6 /nn forlength, 4.9±0.6 (im for width, and 197±26 /im2 forsurface area in the shunted arteries; 79±7 ysn for length,



ownloaded from



JFigure 5. Endothelial intercellular junctions (I) in the canine carotid arteries.A. Shunted artery. B. Control artery. Intercellular junctions In the shunted arteryare long and simple, while those of control arteries are short and irregular.X16000

8.6±1.4 /im for width, and 418±72 /xm2for surface areain the controls.

DiscussionIn 1893, Thoma13 proposed that blood flow velocity was

intimately related to the growth and dilation of bloodvessels. More recently, Kamiya and Togawa14 showedthat the increased blood flow in a canine AV shunt modelresulted in an increase in radius of the carotid arterycorresponding to a wall shear stress level of about dynes/cm2. Zarins et al.15 demonstrated that an increase inradius resulted in the reestablishment of baseline wallshear stress of about 15 dynes/cm2 in the monkey iliacartery 6 months after anastomosis to the iliac vein. Con-versely, Guyton and Hartley16 showed that flow restrictioninhibited the normal increase in carotid artery diameter

during growth in the rat. It has, therefore, been suggestedthat the internal radius of arteries is altered in response tochanges in flow so as to maintain a constant wall shearstress. The mechanisms underlying wall shear stress-dependent arterial growth and enlargement are not fullyunderstood, nor is the sequence of morphologic changesthat correspond to this apparent adaptive process welldefined. In their experiments, Kamiya and Togawa foundthat more than 6 months were required for the caninecarotid artery lumen diameter to increase sufficiently toreestablish the baseline shear stress, but that lumendiameter had not changed by 1 week after AV shunting.Early diameter changes could not be followed as easily inthe dog as in the monkey, because the canine carotidartery diameter changed much less than the monkey iliacartery.15



ownloaded from



Figure 6. Scanning electron microscopic views of the lumen surfaces of thecanine carotid arteries. A. Shunted artery. B. Control artery. Endothelial cells arenarrow and small with protruding surface microvilli in the shunted artery, while theyare flat and rhomboid in the control artery. Microvilli are not prominent In the controlartery. Bar=10 fim. X1000

The observations of Kamiya and Togawa and those ofMasuda et a l . 1 8 1 9 2 0 suggest that immediately before asignificant increase in carotid internal diameter (i.e., at4 weeks) endothelial cells become very active. The pres-ent experiments were, therefore, designed to characterizeand quantify the morphologic changes of endothelial cellsin the flow-loaded canine carotid artery at 4 weeks, theonset of arterial dilation. Since the interaction betweenwall shear stress and the artery wall occurs mainly at theblood-endothelial interface, our objective was to furtherdefine the role of the endothelial cell in shear stress-dependent arterial growth and dilation.

In our experiments, WSS increased sharply immedi-ately after anastomosis and remained elevated during the4-week experimental period on the shunted side. On thecontrol side, WSS was unchanged immediately aftershunting and was only moderately elevated at 4 weeks.This elevation was probably due to the collateral flow

through the circle of Willis from left to right because bloodpressure in the right common carotid artery distal to theAV anastomosis was presumably decreased. WSS on theshunted side was, however, always much higher than onthe control side. Blood flow profiles on the shunted side wereapparently flatter and flow was possibly turbulent, while onthe control side, the Poiseuille flow was maintained.

We found that the endothelial cells in the shuntedarteries were closely packed and radially thickened,protruding markedly into the lumen at 4 weeks. Endo-thelial cell density was elevated, and the abluminaltransverse width of the cells was markedly narrowed.The nuclei on the shunted side were prolate-spheroidbut no greater in volume than those on the control side.Endothelial cell volume per unit of lumen surface areawas about 1.4 times greater on the shunted side than onthe control side, and the volume of the endothelial cellnuclei per unit lumen surface area was 1.8 times larger in



ownloaded from



DFigure 7. Higher magnification of scanning electron microscopic viewsof endothelial cells of shunted arteries show a series of changes thatsuggest mrtotic activity. A. A nuclear constriction. B. A dumbbell-shapednucleus. C and D. Endothelial cells arranged symmetrically with anintervening silt or crevice. Bar=5 nm. x4800



ownloaded from



Table 2. Endothelial Cell Density

Shunted (n=7)

Control (n=7)

Nonoperatedcontrol (n=4)

Light

Lumencircumference

(mm)

8.26 ±0.64

7.91 ±0.91

8.67±1.37

microscopy

Totalnuclei per

section

488±69t||

317±40

277±38

Nuclei/mm

59±6*§

4 1 ± 4 * *

32±2

Electron microscopy

TEM composites

Nuclei/mm

52± 11+11

(n=6)30±4

35±6

Cells/mm

322±22f§(n=6)

224±27"

177±10

SEM

Cells/mm2x10-3

6.15±0.68*§

3.33±0.70tt

2.39±0.46

All values are means±standard deviation. Values on the shunted side are significantly larger than on the controls:*p<0.01 x10"3, tp<0.0001, and +p<0.001 compared to operated controls; §p<0.01 x10"3, ||p<0.001, and Hp<0.05compared to nonoperated controls; **p<0.01, t tp<0.05 compared to nonoperated controls.

Table 3. Dimensions of Endothelial Cells and Nuclei

Cells Nuclei

Length(mm)

Width(Am)

Abluminal area(Mm2)

Volume Length Width Area Volume

Shunted (n=6) 77±13± 7.9±1.1*§ 241±45*§

Control 97±17" 11.2±0.9 432±49tt

Nonoperatedcontrol (n=4) 66±10 12.8+1.0 374±41

331 ±53+ 12.1 ±0.4*11 4.7±0.5t|| 40±2*§ 68±8

430±104tt 13.4±0.5 5.9±0.6 54±4 70±17

306+23 13.0±0.6 6.6+1.0 58±4 60+5

All values are means±standard deviation. Values in the shunted arteries are smaller than in controls: *p<0.001,tp<0.01, and ±p<0.05 compared to operated controls, §p<0.001, ||p<0.01, Hp<0.05 compared to nonoperatedcontrols. Values in operated controls are larger than in nonoperated controls, **p<0.01, t tp<0.05 compared tononoperated controls.

Shunt

Control

Figure 8. Representations of endothelial cells of shunted and control arteries. The upper pair areshunted on the high wall shear stress side. The lower pair are the control side with only a slightincrease in wall shear stress. The drawings are derived from computer-assisted histograms of themorphometric studies obtained from the composite transmission electron micrographic (TEM)photographs. The lower drawings of each pair were obtained by simple smoothing by eye of theblock-like drawings after superimposing scanning electron microscopic and TEM findings.

the shunted arteries. Endothelial cells in the sham-operated control arteries of shunted animals were slightlyenlarged and showed a significant increase in cell densitycompared to nonoperated controls. This difference isprobably attributable to significant hemodynamic changes;the effects of the operative manipulations alone can beexcluded, since there were no morphologic differences inthe right common carotid arteries of operated and nonop-erated controls.

We found several cells with constricted, cleaved, ordouble nuclei, as well as a few twin cell groupings on theshunted side; these features are indicative of mitoticactiv-ity. More direct evidence for proliferation is indicated bydata obtained recently by us in a pilot study with Bromode-oxyuridine (BrdU), a thymidine analogue, which is incor-porated into DNA in the S phase and can be detected bymeans of monoclonal antibodies.2829 Thus far, findingsare available for five adult female beagle dogs with



ownloaded from


822 ARTERIOSCLEROSIS V O L 9, No 6, NOVEMBER/DECEMBER 1989

shunts. After intraperitoneal injection of BrdU (50 mg/kg),we found that 5% to 10% of the endothelial cells on theshunted side had incorporated BrdU at 2 weeks after theoperation, while less than 1% of the cells on the controlside had incorporated BrdU. At 4 weeks, we found that20% of the endothelial cells on the shunted side hadincorporated BrdU compared to less than 1 % on the controlside. A complete 6-month time-course study is underwayand will be published in detail. On the basis of the findingsreported in this communication, we conclude that endothe-lial cells proliferated on the shunted side but were crowdedalong the arterial circumference, which had not yet becomesufficiently enlarged to accommodate the cells in a normalconfiguration. The preliminary data of the BrdU studiesprovide additional support for a proliferative effect.

Although there have as yet been no other reports inwhich endothelial cell density or replication have beenrelated to wall shear stress in vivo, several reports thatsuggest that blood flow affects endothelial cell activityhave appeared. These include local differences in aorticendothelial cell turnover,3031 increased mitotic activity ofaortic endothelial cells about branchings in normal guineapigs,32 and more rapid regeneration of endothelial cells inthe axial direction than about the circumference in the ratafter denudation.33 Increased endothelial cell density after40 days of experimental hypertension and after a sharprise of thymidine index at 7 to 10 days has been observedin the rat aorta.10 Increase of endothelial cell density alsooccurred after experimentally induced stenosis of therabbit aorta,11 as well as increased mitotic activity proxi-mal to experimental aortic coarctatJon in guinea pigs.12

Low shear stress,30 turbulence,32 altered flow velocity,33

and pressure1112 were considered to be the underlyinghemodynamic stimuli in these reports.

In vitro studies have also indicated that endothelial cellsare activated by hemodynamically related stresses, includ-ing elevated wall shear stress.7893435 Ando et al.7 showedthat endothelial migration and proliferation could be elic-ited by increased shear stress with no relationship topressure, pulsation, or chemical growth factors.36 Davieset al.34 demonstrated that turbulent shear stress inducedvascular endothelial turnover, while Levesque and Nerem36

found that, compared to steady-flow shear stress, pulsa-tile shear stress produced an enhanced proliferativeresponse in subconfluent endothelial cell monolayers.These findings tend to indicate that wall shear stresselicits endothelial cell responses under a variety of differ-ences in pattern of shear development (i.e., steady,turbulent, or pulsatile).

Kamiya and Togawa14 suggested that wall shear stress-dependent enlargement is induced by a humoral factorentering the media from the circulation and that the entryof this factor into the artery wall is modulated by wall shearstress. Others have proposed that proliferating endothelialcells have leaky junctions.37 Our demonstration of straightsimple junctions between endothelial cells just before theonset of artery enlargement would tend to support thesuggestion of Kamiya and Togawa, as well as the possi-bility that endothelial cell proliferation results in increasedpermeability due to leaky junctions.

Vasoactive agents, such as endothelial-derived relax-ation factor and endothelial-derived constriction factor,may be released in response to wall shear stress.1738

These agents may also mediate the more chronic type ofwall shear stress-dependent arterial adaptive enlarge-ment that forms the basis of the present study. Anappropriate time-course study of these effects, as well asthe corresponding changes in the media at both thesubacute and chronic phases of the response, shouldhelp to illuminate the mechanisms by which wall shearstress modulates or mediates flow-dependent arterialgrowth and dilation.

Fluid dynamic factors, and wall shear stress in particu-lar, have also been shown to be determinants of plaquelocalization in atherogenesis.38 Endothelial damage wasinduced by markedly elevated experimental wall shearstress,40 and shear stress-related injury has been pro-posed as the basis for plaque localization.41 Zarins et al.42

found, however, that intimal thickening and atherosclero-sis were localized in low shear regions of the humancarotid bifurcation and that increased flow velocity inhib-ited experimental atherogenesis.43 Ross et al.44 havesuggested that endothelial injury, including possible focaldesquamatjons related to altered wall shear stress, mayeventuate in smooth muscle cell proliferation and that thiseffect may be mediated by growth factors derived fromplatelets, endothelial cells, and macrophages. The obser-vations reported in the present report indicate that endo-thelial cells proliferate and change size and configurationin response to increased wall shear stress before theonset of adaptive enlargement. The mechanisms by whichfluid dynamics influence the induction and evolution ofatherosclerosis may be closely related to those thatdetermine vessel enlargement, including the adaptiveenlargement and modelling that accompanies early humanatherogenesis.45 Further investigations of both phenom-ena should eventually provide clinically useful insights intothe adaptive potential of arteries in health and disease.

AcknowledgmentsWe thank Seymour Qlagov, Professor of Pathology at the

University of Chicago, for his advice and reviewing of themanuscript, and Miklo Kobayashi, Telsao Chlda, Nobuyuki Musha,Kenji Nanao, and Yasuhiro Sugihara for their skillful assistancewith the experiments.

References1. Comhlll JF, Levesque MJ, Nerem RM, Kllman JW, Vasko

JS. Quantitative study of the rabbit aortic endothellum usingvascular casts. Atherosclerosis 1980;35:321-327

2. Davies PF, Bowyer DE. Scanning electron microscopy:Arterial endothelial integrity after fixation at physiologicalpressure. Atherosclerosis 1975;21:463-469

3. Flaherty JT, Pierce JE, Ferraris VJ, Patel DJ, Tucker WK,Fry DL. Endothelial nuclear patterns in the canine arterialtree with particular reference to hemodynamic events. CircRes 1972;30:23-33

4. Hlrsch E, MarUno W, Orr CH, White H, Chlsolm GM III.Simple rapid method for the preparation of en face endothe-lial (Hautchen) monolayers from rat and rabbit aortas. Ath-erosclerosis 1980;37:539-548



ownloaded from


ENDOTHEUAL RESPONSE TO INCREASED FLOW Masuda et al. 823

5. Levesque MJ, Cornhlll JF, Nerem RM. Vascular casting. Anew model for the study of the arterial endothellum. Athero-sclerosis 1979:34:457-467

6. Sllkworth JB, Stehbens WE. The shape of endothelial cellsIn en face preparations of rabbit blood vessels. Anglology1975:26:474-487

7. Ando J, Nomura H, Kamlya A. The effect of fluid shearstress on the migration and proliferation of cultured endothe-lial cells. Mterovasc Res 1987;33:61-70

8. Deway CF Jr, Bussolarl SR, Glmbrone MA Jr, Davles PF.The dynamic response of vascular endothelial cells to fluidshear stress. J Biomech Eng 1981:103:177-185

9. Eskln SG, Ives CL, Mclrrtlre LV, Navarro LT. Response ofcultured endothelial cells to steady flow. Microvasc Res1984:28:87-94

10. De Chastonay C, Gabblanl G, Elemer G, Huttner I. Remod-eling of the rat aortic endothelial layer during experimentalhypertension; changes In replication rate, cell density, andsurface morphology. Lab Invest 1983;48:45-52

11. Legg MJ, Gow BS. Scanning electron microscopy of endo-thellum around an experimental stenosis in the rabbit aortausing a new casting material. Atherosclerosis 1982;42:299-318

12. Wright HP. Mitosis patterns in aortic endothelium. Athero-sclerosis 1972:15:93-100

13. Thoma R. Histogenese und Histomechanlk des Gefasssys-tems. Stuttgart: Enke, 1893

14. Kamlya A, Togawa T. Adaptive regulation of wall shearstress to flow change in the canine carotid artery. Am JPhysiol 1980;239:H14-21

15. Zarlns CK, Zatlna MA, Glddens DP, Ku DN, Glagov S.Shear stress regulation of artery lumen diameter in experi-mental atherogenesis. J Vase Surg 1987;5:413-420

16. Guyton JR, Hartley CJ. Row restriction of one carotid arteryin juvenile rats inhibits growth of arterial diameter. Am JPhysiol 1985;248:H540-546

17. Langille BL, O'Donnell F. Reduction in arterial diameterproduced by chronic decreases in blood flow are endothelium-dependent. Science 1986:231:405-407

18. Masuda H, Klkuchl Y, Nemoto T, Bukarl A, Togawa T,Kamlya A. Ultrastructural changes in the endothelial surfaceof the canine carotid artery Induced by wall shear stress load.Biorheology 1982:19:197-208

19. Masuda H, Shozawa T, Kanda M, Kamlya A. Endothelialsurface of the blood flow loaded canine carotid artery. Ascanning and transmission electron microscopical study.Acta Pathol Jpn 1985:35:1037-1046

20. Masuda H, Shozawa T, Hosoda S, Kanda M, Kamlya A.Cytoplasmlc mtarofilaments In endothelial cells of flow loadedcanine carotid arteries. Heart Vessels 1985;1:65-69

21. Masuda H, Salto N, Kawamura K, Sageshlma M, ShozawaT, Kanazawa A. Row loaded canine carotid artery. I. Amorphometric study of microfllament bundles in endothelialcells. Acta Pathol Jpn 1986:36:1833-1842

22. Masuda H, Salto N, Kawamura K, Shozawa T, KanazawaA, Sageshlma M. Row loaded canine carotid artery. II.Ultrastructural changes in the subendothelial layer. ActaPathol Jpn 1987:37:239-251

23. Zarlns CK, Zatlna MA, Glagov S. Correlation of postmor-tem angiography with pathologic anatomy. Quantitation ofatherosclerotic lesions. In: Bond MG, Insull W, Jr, Glagov S,Chandler AB, Cornhill JF, eds. Clinical diagnosis of athero-

sclerosis. Quantitative methods of evaluation. New York:Springer-Verlag, 1986:283-306

24. Waibel ER. Stereological method, vol 2. London: AcademicPress, 1980:175-214

25. Whltmore Rl_ Rheology of the circulation. London: Perga-mon Press, 1968

26. Caro CG, Pedley TJ, Schroter RC, Seed WA. The mechan-ics of the circulation. Oxford: Oxford University Press, 1978

27. Ederstrom HE, de Boer B. Changes in the blood of the dogwith age. Anat Rec 1946:94:663-670

28. Gartmer HG. Monoclonal antibody to 5-bromo- and 5-iododeoxyuridlne: A new reagent for detection of DNA repli-cation. Science 1982;218:474-475

29. Suglhara H, Hattoii T, Fukuda M. Immunohlstochemlcaldetection of bromodeoxyuridine in formalin-fixed tissues.Histochemlstry 1986:85:193-195

30. Caplan BA, Schwartz CJ. Increased endothelial cell turn-over in areas of in vivo Evans blue uptake In pig aortaAtherosclerosis 1973:17:401-417

31. Schwartz SM, Bendrtt EP. Cell replication in the aorticendothelium: A new method for study of the problem. LabInvest 1973:28:699-707

32. Wright HP. Endothelial mitosis around aortic branches Innormal guinea pigs. Nature 1968:220:78-79

33. Schwartz SM, Gajdusek CM, Reldy MA, Selden SC III,Haudenschlld CC. Maintenance of integrity in aortic endo-thelium. Fed Proc 1980:39:2618-2625

34. Davles PF, Remuzzi A, Gordon EJ, Dewey CF Jr, Glm-brone MA Jr. Turbulent fluid shear stress induces vascularendothelial cell turnover In vitro. Proc Natl Acad Sci USA1986:83:2114-2117

35. Levesque MJ, Nerem RM. Pulsatile sheer stress effects oncultured vascular endothelial cells [abstr]. Phys Med Blol1988;33(suppl l):254

36. Jaye M, Howk R, Burgess W, et al. Human endothelial cellgrowth factor: Cloning, nucleotide sequence, and chromo-some localization. Science 1986:233:541-545

37. Welnbaum S, Tzeghal G, Ganatos P, Pfeffer R, Chlen S.Effect of cell turnover and leaky junctions on arterial macro-molecular transport. Am J Physiol 1985;248:H945-960

38. Furchgott RF. Role of endothelium in responses of vascularsmooth muscle. Clrc Res 1983:53:557-573

39. Glagov S, Zarlns CK, Glddens DP, Ku DN. Hemodynamicsand atherosclerosis. Insights and perspectives gained fromstudies of human arteries. Arch Pathol Lab Med1988:112:1018-1031

40. Fry Hl_ Acute vascular endothelial changes associated withincreased blood flow gradients. Circ Res 1968:22:165-197

41. Fry DL. Response of arterial wall to certain physiologicalfactors. In: Atherogenesis: Initiating Factors. Chiba Founda-tion Symposium, Elsvier. Excerpta Medlca Amsterdam: North-Holland, Associated Scientific Publishers, 1973:93-125

42. Zarlns CK, Glddens DP, Bharadva] BK, Sottlural VS,Mabon RF, Glagov S. Carotid bifurcation atherosclerosis.Quantitative correlation of plaque localization with flow veloc-ity profiles and wall shear stress. Circ Res 1983:53:502-514

43. Zarlns CK, Bomberger RA, Glagov S. Local effects ofstenosis: Increased flow velocity inhibits atherogenesis. Cir-culation 1981;64(suppl ll):221-227

44. Ross R, Glomset J, Harker L_ Response to injury andatherogenesis. Am J Pathol 1977;86:675-684

45. Glagov S, Welsenberg E, Zarlns CK, Stankunavldua R,Kolettis G. Compensatory enlargement of human atheroscle-rotic coronary arteries. N Engl J Med 1987;316:1371-1375

Index Terms: endothelial cell • blood flow •• morphometry

canine carotid arteryultrastructure

arteriovenous shunt



ownloaded from


H Masuda, K Kawamura, K Tohda, T Shozawa, M Sageshima and A Kamiyaartery.

Increase in endothelial cell density before artery enlargement in flow-loaded canine carotid

Print ISSN: 1079-5642. Online ISSN: 1524-4636 Copyright © 1989 American Heart Association, Inc. All rights reserved.

Avenue, Dallas, TX 75231is published by the American Heart Association, 7272 GreenvilleArteriosclerosis, Thrombosis, and Vascular Biology

doi: 10.1161/01.ATV.9.6.8121989;9:812-823Arterioscler Thromb Vasc Biol.

http://atvb.ahajournals.org/content/9/6/812World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://atvb.ahajournals.org//subscriptions/

at: is onlineArteriosclerosis, Thrombosis, and Vascular Biology Information about subscribing to Subscriptions:

http://www.lww.com/reprints

Information about reprints can be found online at: Reprints:

document.Permissions and Rights Question and AnswerFurther information about this process is available in theis being requested is located, click Request Permissions in the middle column of the Web page under Services.Clearance Center, not the Editorial Office. Once the online version of the published article for which permission

can be obtained via RightsLink, a service of the CopyrightArteriosclerosis, Thrombosis, and Vascular Biology Requests for permissions to reproduce figures, tables, or portions of articles originally published inPermissions:



ownloaded from

http://atvb.ahajournals.org/content/9/6/812

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://atvb.ahajournals.org//subscriptions/


increase in endothelial cell density before artery ... · pdf fileshunt was closed after the...

Documents