Endothelial nitric oxide synthase affects both earlyand late collateral arterial adaptation and bloodflow recovery after induction of hind limbischemia in miceBrian Park, MD,a Ari Hoffman, MD,b Yagai Yang, PhD,b Jinglian Yan, PhD,a Guodong Tie, PhD,a

Hossein Bagshahi, MD,a Philip T. Nowicki, MD,a and Louis M. Messina, MD,a Worcester, Mass; and SanFrancisco, Calif

Objective: The goals of this study were to determine if endothelial nitric oxide synthase (eNOS) affects both early and latecollateral arterial adaptation and blood flow recovery after severe limb ischemia in a mouse model and to determine ifeNOS-derived NO is necessary for recruitment of chemokine (C-X-C motif) receptor 4 (CXCR4)� vascular endothelialgrowth factor receptor-1 (VEGFR1)� hemangiocytes to the site of ischemia.Methods: Two studies were completed. In the first, hind limb ischemia was induced by unilateral femoral artery excisionin three groups: C57Bl6 (wild-type), eNOS�/�, and C57Bl/6 mice treated with NG-nitro-L-arginine methyl ester(L-NAME) from 1 day before excision through day 3 after excision (early L-NAME group). These groups were studiedon day 3 after induction of ischemia. In the second study, hind limb ischemia was induced in C57Bl/6 mice (wild-type)and C57Bl/6 mice treated with L-NAME from days 3 through 28 after induction of ischemia. These groups were studiedday 28 after ischemia induction. Dependent variables included hind limb perfusion, collateral artery diameter, and thenumber and location of hemangiocytes within the ischemic hind limb.Results: In the first study, toe gangrene developed in the eNOS�/� and early L-NAME treatment groups by day 2. Thesegroups demonstrated less blood flow recovery and smaller collateral artery diameter than the wild-type group.Hemangiocytes were present within the adventitia of collateral arteries in the wild-type group but were only sparselypresent, in a random pattern, in the eNOS�/� and early L-NAME treatment groups. In the second study, the lateL-NAME group showed less blood flow recovery and smaller collateral artery diameter on day 28 of ischemia than thewild-type group. Hemangiocytes were present in a pericapillary distribution in the wild-type group, but were present onlysparsely in the late L-NAME treatment group.Conclusion: Early (day 3) and late (day 28) adaptive responses to hind limb ischemia both require eNOS–derived NO. NOis necessary for normal hemangiocyte recruitment to the ischemic tissue. (J Vasc Surg 2010;51:165-73.)

Clinical Relevance: This study demonstrates that endothelial nitric oxide synthase (eNOS)-derived NO is requisite for boththe early and late vascular recovery phases in response to hind limb ischemia. Moreover, it demonstrates that recruitment ofhemangiocytes, a specific subset of vascular progenitor cells that display vascular endothelial growth factor receptor 1(VEGFR1) and chemokine (C-X-C motif) receptor 4 (CXCR4) surface antigens, is dependent on the presence of NO. Thesefindings enhance understanding of the basic biologic mechanisms in the recovery from ischemia. The need for eNOS-derivedNO suggests that clinical strategies to enhance postocclusive flow recovery in peripheral artery disease might be more successfulif eNOS or NO, or both, are applied as a part of the treatment paradigm. Hemangiocytes are an important part of the cellularresponse to ischemia. The present findings underscore their importance and suggest that preservation or enhancement of

eNOS within the ischemic limb might improve the recruitment of these critical reparative cells to the site of ischemic injury.

Critical limb ischemia secondary to peripheral arterialdisease is a debilitating and potentially lethal disease. Crit-ical limb ischemia is responsible for �150,000 amputa-tions each year in the United States.1 In many patients,

From the Division of Vascular Surgery, University of MassachusettsMedical School, Worcestera; and the University of California at SanFrancisco, San Francisco.b

This work was funded by HL75353 to Dr Messina.Competition of interest: none.Reprint requests: Louis M. Messina, MD, University of Massachusetts

Medical School, 55 Lake Ave N, Worcester, MA 01655 (e-mail:MessinaL@ummhc.org).

The editors and reviewers of this article have no relevant financial relationshipsto disclose per the JVS policy that requires reviewers to decline review of anymanuscript for which they may have a competition of interest.

0741-5214/$36.00Copyright © 2010 by the Society for Vascular Surgery.

doi:10.1016/j.jvs.2009.08.045

surgical or catheter-based revascularization proceduresare not possible, and amputations are necessary. For thesereasons, the development of molecular or cell-based therapiesto enhance collateral artery enlargement and angiogenesiscontinues to be an area of substantial scientific and clinicalinterest.

Enlargement of the diameter of existing collateral arteriesis the means by which blood flow downstream from the site ofconduit artery occlusion is restored.2,3 At least three mecha-nisms participate in this process: shear stress,4,5 inflamma-tion,6 and recruitment of bone marrow-derived vascular pro-genitor cells to areas of ischemia.7-10 Nitric oxide (NO) is acomponent in each of these mechanisms.11-14

During the early adaptive response to ischemia, in-creased shear stress within collateral arteries leads to NO-

induced vasodilation that results in a transient, temporary

165

JOURNAL OF VASCULAR SURGERYJanuary 2010166 Park et al

increase in flow through the collaterals.11 During the lateadaptive response to ischemia, NO participates in the re-cruitment of vascular progenitor cells that generate collat-eral artery remodeling, a process termed arteriogenesis,wherein net conductance across the collateral vessels ispermanently increased.12-14

It is not universally accepted that NO derived from theendothelial isoform of nitric oxide synthase (eNOS) isrequisite for arteriogenesis.11 In addition, the involvementof NO in the recruitment of hemangiocytes, a criticallyimportant subset of vascular progenitor cells,10,15 has notbeen defined. This study was designed to determine ifeNOS-derived NO is essential for both the early and lateadaptive response to ischemia—to arteriogenesis—and todetermine the importance of NO in hemangiocyte recruit-ment. We hypothesized that (1) eNOS-derived NO pro-duction is essential for collateral artery remodeling, orarteriogenesis, during the late adaptive response to isch-emia and (2) recruitment of bone marrow-derived heman-giocytes to the ischemic hind limb vasculature requires thepresence of eNOS-derived NO.

MATERIALS AND METHODS

Experimental animals and procedures. The care ofmice used in this study complied with the National Re-search Council’s Guide for the Care and the Use of Labora-tory Animals. All protocols were approved by the Institu-tional Animal Care and Use Committee at the University ofMassachusetts (Worcester, Mass).

C57Bl/6 mice and eNOS�/� mice (Jackson Labora-tories, Bar Harbor, Me) were fed standard chow and main-tained on a 12-hour light/dark cycle. Hind limb ischemiawas generated by ligation of the femoral artery at theinguinal ligament and popliteal fossa, followed by excisionof the artery and all branches.16 The procedure was doneunder 3% isoflurane anesthesia. Buprenorphine was admin-istered during the immediate postoperative period.

Experimental protocols. The early and late adaptiveresponses to acute hind limb ischemia were evaluated inseparate studies. The early adaptive response to ischemia(vasodilation) was studied 3 days after unilateral femoralartery excision in three groups: C57Bl/6 mice (wild-typegroup), mice with targeted disruption of the eNOS loci(eNOS�/� group), and C57Bl/6 mice treated with NG-nitro-L-arginine methyl ester (L-NAME; 1 g/L) added tothe drinking water 24 hours before femoral artery excision(early L-NAME group). The L-NAME treatment was con-tinued until sacrifice on day 3 after induction of ischemia, atime selected because preliminary work indicated that micein the eNOS�/� and early L-NAME groups uniformlydeveloped limb gangrene by this time.

The late adaptive response to ischemia (arteriogenesis)was studied 28 days after induction of ischemia in twogroups: C57Bl/6 mice (wild-type group) and C57Bl/6mice treated with L-NAME (1 g/L in drinking water)beginning on day 3 after the induction of hind limb isch-emia (late L-NAME group). The L-NAME treatment was

continued until sacrifice. Day 28 after induction of hind

limb ischemia was chosen as the time for study in the lateadaptive response experiment because published reportsindicate that postischemia arteriogenesis is evident at thattime.3 The late adaptive response to ischemia was notstudied in eNOS�/� mice because significant autoamputa-tion consistently developed in these animals consistently byday 3 after induction of ischemia and we did not believe itwas ethically appropriate to keep them alive �3 days.

Laser Doppler perfusion imaging. A laser Dopplerperfusion imager (LDPI; Moor Instruments Ltd, Devon,United Kingdom) was used to estimate blood flow withincalf muscles. LDPI-derived data strongly correlate with calfmuscle blood flow measured directly using microspheres,and the method is widely used to quantify perfusion in therodent hind limb.16

Hair from limbs was removed by depilatory cream, andmice were placed on a heating plate at 37°C to minimizetemperature variation. Studies were conducted under 1.5%isoflurane anesthesia. Data are presented as the ratio ofperfusion to the ischemic hind limb and the contralateralnonischemic hind limb.

Immunostaining of thigh collateral arteries. Thethigh muscles were harvested on day 3 or 28 after inductionof ischemia in the early or late adaptive response groups,respectively. Collateral arteries were identified by doublestaining with antibodies for CD31 (platelet endothelial celladhesion molecule-1, a marker of endothelial cells) andsmooth muscle actin (both antibodies from BD Bio-sciences, San Jose, Calif). Collateral artery diameter wasmeasured in 10 randomly selected low-power fields usingprecalibrated microscope calipers (Carl Zeiss, Göttingen,Germany), and the average diameter for the ischemic andcontralateral nonischemic hind limb was determined foreach mouse. Collateral arterial diameter was expressed as aratio of the diameters in ischemic and nonischemic hindlimbs. All measurements were made in a blinded manner.

Immunostaining for hemangiocytes. The presenceof hemangiocytes was measured in thigh muscles on day 3or 28 after the induction of ischemia in the early and lateadaptive response groups, respectively. Hemangiocyteswere identified by double staining for chemokine (C-X-Cmotif) receptor 4 (CXCR4) and vascular endothelialgrowth factor receptor 1 (VEGFR1; both antibodies fromBD Bioscience).10 The number of hemangiocytes was de-termined in 10 randomly selected high-power fields inischemic and contralateral and nonischemic thigh muscle,and the average was calculated. Observations were made ina blinded manner.

Confocal microscopy. A Leica DM IRBE confocal mi-croscope was used (Leica Microsystems, Wetzler, Germany).Cy3-conjugated CXCR4 and fluorescein isothiocyanate-conjugated VEGFR1 antibodies (BD Bioscience) wereused to identify hemangiocytes. Nuclei were identified byDraq5 staining (Biostatus Limited, Leicestershire, UnitedKingdom).

Nitrate/nitrite assay. Serum samples were processedusing a nitrate/nitrite (NOx) fluorometric assay kit (Cay-

man Chemical, Ann Arbor, Mich), which determines the

JOURNAL OF VASCULAR SURGERYVolume 51, Number 1 Park et al 167

concentration of NO2– and NO3

–, the final products gen-erated during NO metabolism in vivo. Fluorescence wasdetected using a Biotek Synergy 2 detector (BioTek Instru-ments Inc, Winooski, Vt).

Statistical analysis. Differences among groups at eachtime point were analyzed by analysis of variance (ANOVA)for comparisons between three groups of mice, and the ttest for comparisons between two groups, where appropri-ate. If the ANOVA f statistic was significant (P � .05), thenpost hoc Student-Newman-Keuls tests were done to deter-mine the sites of significance at the P � .05 level threshold.

RESULTS

Early adaptive response to hind limb ischemia. Theserum NOx concentration in the eNOS�/� group was 67%less than that noted in the wild-type group. Treatment ofC57Bl/6 mice with L-NAME reduced serum NOx by 71%on day 1 and by 66% on day 3. These levels were similar tothose noted in the eNOS�/� group (Fig 1).

The eNOS�/� and early L-NAME treatment groupsdemonstrated significant toe necrosis by the second dayafter induction of ischemia, whereas wild-type mice did not(Fig 2, A). All three groups demonstrated a significantreduction of flow to the ischemic hind limb immediatelyafter femoral artery excision, whereas flow to the contralat-eral hind limb did not change; consequently, the ischemic/nonischemic limb flow ratio decreased in all three groups.This circumstance remained unchanged in all three groupsthrough day 2. On day 3, however, the wild-type groupdemonstrated an increase in flow to the ischemic hind limb,resulting in an ischemic/nonischemic flow ratio that wasmore than twofold greater than that noted in theeNOS�/� or L-NAME groups (Fig 2, B). The ischemic/nonischemic collateral artery diameter ratio was 45% lower

Fig 1. Serum nitrite (NOx) concentration in wild-type, endothe-lial nitric oxide synthase (eNOS�/�) and early NG-nitro-L-argininemethyl ester (L-NAME) treatment groups. The serum NOx levelwas lower in the eNOS�/� mice than in the wild-type mice.L-NAME significantly reduced serum NOx levels �1 day afterbeginning L-NAME treatment, and the reduction was sustainedthroughout the duration of L-NAME treatment.

in the eNOS�/� group and 55% lower in the early L-

Fig 2. Responses of wild-type, endothelial nitric oxide synthase(eNOS�/�) and early NG-nitro-L-arginine methyl ester (L-NAME)treatment groups to hind limb ischemia. A, Effects of femoral arteryexcision on hind limb tissue integrity are shown in photographs that arerepresentative of all animals in each group. The ischemic hind limb isshown on the left and white arrows in eNOS�/� and L-NAME micedenote evidence of tissue necrosis. B, Laser Doppler perfusion imager(LDPI) data are shown. Each mouse underwent 3 scans at each timepoint, and the ratio of the LDPI signal from the ischemic and nonisch-emic hind limb was calculated; the average ratio was used as a single datapoint for each mouse. C, Collateral artery diameter on day 3 after theinduction of hind limb ischemia was determined as the average of thelargest arteries identified within 10 randomly selected low-power fieldsfor each mouse by a blinded observer. The ratio of the average diameterin the ischemic thigh to the average diameter in the nonischemic thigh

was determined for each mouse and used as a single data point.

JOURNAL OF VASCULAR SURGERYJanuary 2010168 Park et al

Fig 3. Presence of hemangiocytes in the thigh musculature in wild-type, endothelial nitric oxide synthase (eNOS�/�),and early NG-nitro-L-arginine methyl ester (L-NAME) treatment groups on day 3 after induction of hind limbischemia. A, Quantitation of hemangiocytes, defined as cells that colocalized chemokine (C-X-C motif) receptor 4(CXCR4) and vascular endothelial growth factor receptor (VEGFR) 1, was determined as the average number ofhemangiocytes present in each mouse in 10 randomly selected high-power fields, determined by a blinded observer.B, In representative photomicrographs (original magnification �200) of hemangiocyte immunostaining, VEGFR1 isstained green, CXCR4 is stained red, and nuclei are stained blue. A collateral vessel is present in each photomicrograph,evidenced by the linear cord of blue-stained nuclei. Note the presence of double-stained hemangiocytes clustered alongthe collateral artery in the ischemic thigh of the control mouse. C, Confocal microscopy of a collateral artery within theischemic thigh of the wild-type group. The left photo is shown at original magnification �200. The area delineated bythe white square in the left photo is shown at original magnification �400 in the right photo. Note the presence of ahemangiocyte in the adventitial area of a collateral artery. This adventitial distribution of hemangiocytes was commonlyobserved in the ischemic thigh muscle of wild-type mice, but it was never observed in eNOS�/� or early L-NAME

treatment groups.

used

JOURNAL OF VASCULAR SURGERYVolume 51, Number 1 Park et al 169

NAME group than in the wild-type group on day 3(Fig 2, C).

The number of CXCR4� VEGFR1� hemangiocyteswithin the ischemic thigh muscle was 86% less in theeNOS�/� and early L-NAME treatment groups than inthe wild-type group (Fig 3, A). These cells were clearlyaligned along collateral arteries in the ischemic thighmusculature in the wild-type group (Fig 3, B). Confo-cal microscopy revealed that hemangiocytes were pri-marily present within the adventitia of the collateralarteries within the ischemic thigh in the wild-type group(Fig 3, C).

Late adaptive response to hind limb ischemia.L-NAME treatment in the late adaptive response experi-ment was initiated on the third day after induction of hindlimb ischemia, after the LDPI measurement was obtained.Blood flows to the ischemic and nonischemic hind limbswere similar in the wild-type and late L-NAME treatmentgroups through day 7. Beginning on day 14, blood flow tothe ischemic hind limb became significantly greater in thewild-type group than the late L-NAME group, whereasflows to the contralateral nonischemic hind limb wereunchanged; accordingly, the ischemic/nonischemic flowratios in the L-NAME group were 33% and 44% less than of

Fig 4. Response of wild-type and late NG-nitro-L-arginischemia. A, Laser Doppler perfusion imager (LDPI) datratio of the LDPI signal from the ischemic and nonischesingle data point for each mouse. B, Collateral artery ddetermined as the average of the largest arteries identifiedby a blinded observer. The ratio of the average diamnonischemic thigh was determined for each mouse and

those in the wild-type group on days 14 and 28, respec-

tively (Fig 4, A). The ischemic/nonischemic thigh collat-eral diameter ratio was 60% less in the late L-NAME groupthan the wild-type group on day 28 after induction of hindlimb ischemia (Fig 4, B).

The number of CXCR4� VEGFR1� hemangiocytespresent within the ischemic thigh muscles was 56% less inlate L-NAME treatment group than in the wild-type groupon day 28 after induction of hind limb ischemia, whereasthe number of hemangiocytes present within the contralat-eral nonischemic thigh muscles was similar in both groups(Fig 5, A). Interestingly, the distribution of these cells wasdifferent than that observed in the early adaptive responsestudy on day 3 after induction of hind limb ischemia. Thus,hemangiocytes were present between myofibers at the sitesoccupied by capillaries (Fig 5, B). Confocal microscopyconfirmed the pericapillary location of these hemangiocytes(Fig 5, C).

DISCUSSION

Adaptations within existing collateral arteries occurafter femoral artery occlusion to restore vascular conduc-tance and, hence, limb blood flow. The early adaptation isvasodilation, whereas the late adaption is arteriogenesis, thelater being a process of collateral artery remodeling.3 Re-

ethyl ester (L-NAME) treatment groups to hind limbch mouse underwent 3 scans at each time point, and theind limb was calculated; the average ratio was used as a

er on day 28 after induction of hind limb ischemia wasch mouse within 10 randomly selected low-power fieldsin the ischemic thigh to the average diameter in theas a single data point.

ine ma. Eamic h

iametin ea

eter

search has established that eNOS-derived NO mediates

JOURNAL OF VASCULAR SURGERYJanuary 2010170 Park et al

JOURNAL OF VASCULAR SURGERYVolume 51, Number 1 Park et al 171

collateral artery vasodilation during the early adaptive re-sponse to ischemia; however, the role of eNOS-derived NOin arteriogenesis remains unclear.11,17-21 The present stud-ies support a role for eNOS-derived NO in both early andlate collateral artery adaptations. In addition, this studyprovides the first evidence, to our knowledge, that eNOS isrequisite for recruitment of hemangiocytes, a unique subsetof bone marrow-derived vascular progenitor cells,10,15 tothe ischemic vasculature.

Occlusion of the femoral artery increases the rate ofblood flow through pre-existing, small-diameter collateralarteries. This circumstance increases wall shear stress inthese vessels, a mechanical stimulus that increases eNOSexpression and activity, the latter through eNOS phosphor-ylation.5,17 These events significantly enhance NO produc-tion and induce vasodilation of existing collateral arteries;new collateral arteries are not created during this process ofarteriogenesis.3,5,17 The present findings are fully consis-tent with this process, as collateral artery diameter andperfusion within the ischemic hind limb were less on day 3after induction of ischemia in the eNOS�/� and earlyL-NAME treatment groups than in the wild-type group.

It is known that the early postischemic, NO-inducedvasodilation of collateral arteries is short-lived.3,11,18 Shearstress, the primary stimulus for this event, dissipates duringvasodilation because shear stress is inversely proportional tothe fourth power of the vessel radius.18 This temporaryvasodilation gives way to collateral artery remodeling, orarteriogenesis, a process that permanently increases netcollateral artery conductance.2,3 Heretofore, it was notclear if eNOS-derived NO participates in the process ofarteriogenesis. Yu et al19 and Lloyd et al20 demonstratedreduced postischemic arteriogenesis when eNOS activitywas blocked or eliminated; in contrast, Mees et al11 failedto demonstrate an effect of eNOS-derived NO beyond 2weeks after induction of ischemia. The present findingssupport a role for NO during arteriogenesis, because col-lateral artery diameter and blood flow within the ischemichind limb were significantly less in the late L-NAME treat-ment group than in the wild-type group 28 days after theinduction of ischemia. Collectively, our findings indicatethat NO is an active participant in postischemic vascularadaptation for at least the first month after the induction ofischemia.

4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™Fig 5. Presence of hemangiocytes in the thigh muscu(L-NAME) treatment groups 28 days after induction of has cells that colocalized chemokine (C-X-C motif) recepto(VEGFR) 1, was determined as the average number of hhigh-power fields, determined by a blinded observerimmunostaining, VEGFR1 is stained green, CXCR4 is stobtained at 3 days after the induction of hind limb ischembetween myocytes; that is, at the sites of capillaries (orischemic thigh of the wild-type group. The left-handdelineated by the white square in the left photo is shown

presence of hemangiocytes in the pericapillary region.

An alternative means to evaluate the role of eNOS inpostischemic arteriogenesis is by enhancing eNOS activity,a circumstance that might be predicted to increase collat-eral artery growth and blood flow recovery. This predica-tion has been confirmed: the administration of adenovirusvectors encoding eNOS to the ischemic hind limb,21 over-expression of constitutively active eNOS in the ischemichind limb,11 or dietary supplementation with L-arginine17

significantly improved postischemic arteriogenesis. Restitu-tion of eNOS activity or administration of exogenous NOin the late L-NAME treatment group could provideadditional support for the experimental hypothesis, if itrestored arteriogenesis. However, coadministration ofL-arginine and an arginine analogue such as L-NAMEunder in vivo conditions, or systemic administrationof NO donor agents such as sodium nitroprusside,S-nitrosoglutathione, or 3-morpholinosydnonimine, could alsogenerate confounding results because it would not be clearif these agents reached the ischemic hind limb.

Femoral artery excision establishes an inflammatory statewithin the ischemic tissue; monocytic6 and lymphocytic22

infiltration are requisite for postischemic arteriogenesis andangiogenesis. Activity of the Ca2�-independent isoformof nitric oxide synthase (iNOS) is increased in homoge-nates of ischemic hind limb muscle prepared 14 daysafter induction of ischemia; protein and messenger RNAexpression of iNOS were not determined, nor was thespecific localization of the iNOS enzyme within thetissue (personal communication, Jinglian Yan, Worces-ter, Mass, 2009).

The arginine analogue L-NAME interacts with thearginine binding sites of both eNOS and iNOS and, al-though then half maximal inhibitory concentration forL-NAME for each enzyme (3.1 �M for iNOS, 0.35 �M foreNOS), indicate a tenfold selectivity of eNOS � iNOS,24 itis likely that some L-NAME–induced inhibition of iNOSoccurred in this study. In contemplating a possible role foriNOS-derived NO in arteriogenesis, consideration must begiven to the site of NO production, as the exceedingly briefhalf-life of NO precludes a direct effect at a distance farremoved from its point of origin. Additional studies of iNOSare warranted, with specific attention to the location andcellular source of iNOS expression within the ischemic hindlimb; moreover, iNOS inhibition with 1400W, which has

™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™in control and late NG-nitro-L-arginine methyl ester

mb ischemia. A, Quantitation of hemangiocytes, definedXCR4) and vascular endothelial growth factor receptor

giocytes present in each mouse in 10 randomly selectedIn representative photomicrographs of hemangiocytered, and nuclei are stained blue. In contrast to the data

ig 2, B), note that hemangiocytes are present in the areasmagnification �200). C, Confocal microscopy of the

o is shown at original magnification �200. The areaiginal magnification �400 in the right photo. Note the

™™latureind lir 4 (C

eman. B,ainedia (F

iginalphotat or

JOURNAL OF VASCULAR SURGERYJanuary 2010172 Park et al

about a 105-fold specificity for iNOS � eNOS,23 might clarifythe putative role of iNOS-derived NO in arteriogenesis.

An important and novel finding from the present ex-periments is that eNOS-derived NO is requisite for recruit-ment of hemangiocytes to the site of ischemic injury.Hemangiocytes, first described by Jin et al,10 represent aunique subset of bone marrow-derived progenitor cells thatparticipate in postischemic vascular repair.10,15 Hemangio-cytes differ from endothelial progenitor cells based onsurface antigens; thus, hemangiocytes express CXCR4 andVEGFR1, whereas endothelial progenitor cells expressCD31, CD133, and VEGFR2.7,8 It is well established thateNOS-derived NO participates in the recruitment of endo-thelial progenitor cells to the site of ischemic vascularinjury.12-14

The present findings indicate a similar interaction forhemangiocytes. Specifically, hemangiocytes were evident inthe adventitia of collateral arteries in the wild-type group byday 3 after induction of hind limb ischemia, whereas thesecells were present only sporadically within the ischemicthigh when eNOS-derived NO was not present, that is, inthe eNOS�/� or early L-NAME treatment groups. Mobi-lization and homing of CXCR4 hemangiocytes is alsocontingent on the chemokine stromal-derived factor-1�(SDF-1�),10 and an interaction between eNOS-derivedNO and SDF-1� has been described;13 in this context,measurement of SDF-1� in the ischemic hind limb ofeNOS�/� mice and the use of CXCR4 blocking antibodiesin this group represent two potentially useful lines of fur-ther investigation.

Another novel finding from these experiments is thechange in location of hemangiocytes during the course ofpostischemic vascular recovery: on day 3, these cells werepresent along collateral arteries, the site anticipated wherethese cells would be participating in arteriogenesis, whereason day 28, these cells were present in a pericapillary loca-tion. Angiogenesis, defined as the de novo generation ofnew capillaries, is an essential part of postischemic vascularadaptation.3 The capillary/myofiber ratio, a marker forangiogenesis, was not measured in this study; however, thepresence of hemangiocytes in the pericapillary regions be-tween muscle cells might indicate that angiogenesis hadoccurred. Moreover, we propose that NO is requisite forrecruitment of hemangiocytes to the per-capillary location,as evidenced by the significant reduction of pericapillaryhemangiocytes in the late L-NAME treatment group.

Most of the hemangiocytes seen within capillaries exhib-ited fragmentation of their nuclei, an observation that sug-gests apoptosis. It has been proposed that progenitor cellsserve a paracrine effect; thus, these cells affect existing vascularcells, inducing repair and replication, rather than becomingpermanently incorporated into existing vascular tissue.26

Interestingly, gangrene of the toes consistently oc-curred in eNOS�/� and early L-NAME–treated mice byday 3 after induction of hind limb ischemia but was neverobserved in late L-NAME–treated animals. One explana-tion for this finding is that the delay in initiating L-NAME

treatment until day 3 after ischemia allowed sufficient time for

NO-induced recruitment of hemangiocytes to pre-existingcollateral arteries in a manner similar to that observed in thewild-type group studied on day 3. These cells have the poten-tial to restructure the ischemic vasculature by direct12-14 andparacrine24 actions, improving vascular conductance andhence perfusion. Another explanation is that mice in theeNOS�/� and early L-NAME treatment groups were unableto generate the immediate eNOS-derived vasodilation of col-lateral arteries, a circumstance that would lead to a profoundcompromise in downstream perfusion.

We recognize two limitations to the interpretation of ourresults. First, the technique of femoral artery excision pro-duces a sudden and very severe level of ischemia, whereas thedisease process of critical limb ischemia in humans developsover a period of years.1 A more gradual, clinically duplicativemodel of femoral arterial occlusion has been described in ratsthat causes less inflammation and less tissue necrosis but lowerlevels of blood flow recovery than after acute occlusion.16 Thismodel uses an ameroid constrictor, however, which is toolarge for use in the mouse hind limb.

Second, we did not use perfusion fixation at an in situpressure during preparation of the thigh muscles beforecollateral artery diameter measurement; instead, this tissuewas prepared at an effective pressure of 0 mm Hg. Thisapproach certainly limited the diameter of collateral arteries atthe time of measurement. The difference in collateral arterydiameter among study groups was quite substantial, however,and collateral wall thickness did not vary among groups. It isthus unlikely that the use of perfusion fixation would havealtered the outcome of collateral artery diameter.

CONCLUSIONS

The present findings are relevant to human peripheralarterial disease (PAD). PAD patients who develop intermit-tent claudication or critical limb ischemia have been shownto have endothelial dysfunction characterized by decrea-sed NO bioavailability.25 This reduced NO bioavailabilitylikely results in inadequate collateral artery enlargement inPAD. In addition, reduced NO bioavailability may impedethe efficacy of molecular or chemical therapeutics in PAD.Clinical trials administering either recombinant VEGF26 orrecombinant fibroblast growth factor-227 to PAD patientshave demonstrated therapeutic efficacy only after 60 to 90days of treatment. VEGF and recombinant fibroblastgrowth factor-2 both exert their effects by upregulationand activation of eNOS.28,29 The present study indicatesthat eNOS-derived NO is requisite for acute adaptation tosevere ischemia to prevent tissue necrosis and for sustainedarteriogenesis. We propose that treatment with agentsknown to enhance eNOS activity may prove a useful ad-junctive therapy in PAD. Only through a complete under-standing of the specific and temporally distinct eNOS-related recovery responses can we hope to achieve effectivemolecular and cell-based therapies for human patients.

We thank Jianming Li and Phong Dargon for assistance

in obtaining confocal microscopy images.

JOURNAL OF VASCULAR SURGERYVolume 51, Number 1 Park et al 173

AUTHOR CONTRIBUTIONS

Conception and design: BP, LMAnalysis and interpretation: BP, LMData collection: BP, AH, YY, JY, GT, HBWriting the article: BP, PNCritical revision of the article: BP, PN, LMFinal approval of the article: LMStatistical analysis: BPObtained funding: LMOverall responsibility: LM

REFERENCES

1. Schainfeld RM, Isner JM. Critical limb ischemia: nothing to give at theoffice? Ann Intern Med 1999;130:442-4.

2. Helisch A, Schaper W. Arteriogenesis: the development and growth ofcollateral arteries. Microcirculation 2003;10:83-97.

3. Scholz D, Ziegelhoeffer T, Helisch A, Wagner S, Friedrich C, Podzu-weit T, Schaper W. Contribution of arteriogenesis and angiogenesis topostocclusive hindlimb perfusion in mice. J Mol Cell Cardiol 2002;34:775-87.

4. Cooke JP. Flow, NO, and atherogenesis. Proc Natl Acad Sci 2003;100:768-70.

5. Pipp F, Boehm S, Cai WJ, Adili F, Ziegler B, Karanovic G, et al. Elevatedfluid shear stress enhances postocclusive collateral artery growth andgene expression in the pig hindlimb. Arterioscler Thromb Vasc Biol2004;24:1664-8.

6. Voskuil M, van Royen N, Hoefer IE, Seidler R, Guth BD, Bode C, et al.Modulation of collateral artery growth in a porcine hindlimb ligationmodel using MCP-1. Am J Physiol Heart Circ Physiol 2003;284:H1422-8.

7. Asahara T, Murohara T, Sullivan A, Silver M, van der Zee R, Li T, et al.Isolation of putative progenitor endothelial cells for angiogenesis. Sci-ence 1997;275:964-7.

8. Urbich C, Dimmeler S. Endothelial progenitor cells: characterizationand role in vascular biology. Circ Res 2004;95:343-53.

9. Kinnaird T, Stabile E, Burnett MS, Lee CW, Barr S, Fuchs S, et al.Marrow-derived stromal cells express genes encoding a broad spectrumof arteriogenic cytokines and promote in vitro and in vivo arteriogenesisthrough paracrine mechanisms. Circ Res 2004;94:678-85.

10. Jin DK, Shido K, Kopp HG, Petit I, Shmelkov SV, Young LM, et al.Cytokine-mediated deployment of SDF-1 induces revascularizationthrough recruitment of CXCR4� hemangiocytes. Nat Med 2006;12:557-67.

11. Mees B, Wagner S, Ninci E, Tribulova S, Martin S, van Haperen R, et al.Endothelial nitric oxide synthase activity is essential for vasodilationduring blood flow recovery but not for arteriogenesis. ArteriosclerThromb Vasc Biol 2007;27:1-8.

12. Aicher A, Heeschen C, Mildner-Rihm C, Urbich C, Ihling C, Technau-Ihling K, et al. Essential role of endothelial nitric oxide synthase formobilization of stem and progenitor cells. Nat Mede 2003;9:1370-6.

13. Kaminski A, Ma N, Donndorf P, Lindenblatt N, Feldmeier F, Ong L, etal. Endothelial NOS is required for SDF-1�/CXCR4-mediated periph-eral endothelial adhesion of c-Kit� bone marrow stem cells. Lab Invest

2008;88:58-69.

14. Rafii D, Psaila B, Butler J, Jin DK, Lyden D. Regulation of vasculogen-esis by platelet-mediated recruitment of bone marrow-derived cells.Arterioscler Thromb Vasc Biol 2008;28:217-22.

15. Wragg A, Mellad J, Beltran L, Konoplyannikov M, San H, Boozer S, etal. VEGFR1/CXCR4-positive progenitor cells modulate local inflam-mation and augment tissue perfusion by a SDF-1-dependent mecha-nism. J Mol Med 2008;86:1221-32.

16. Tang GL, Chang DS, Sarkar R, Wang R, Messina LM. The effect ofgradual or acute arterial occlusion on skeletal muscle blood flow,arteriogenesis, and inflammation in rat hindlimb ischemia. J Vasc Surg2005;41:312-20.

17. Murohara T, Asahara T, Silver M, Bauters C, Masuda H, Kalka C, et al.Nitric oxide synthase modulates angiogenesis in response to tissueischemia. J Clin Invest 1998;101:2567-8.

18. Eitenmuller I, Volger O, Kluge A, Troidl K, Barancik M, Cai WJ, et al.The range of adaptation by collateral vessels after femoral artery occlu-sion. Circ Res 2006;99:656-62.

19. Yu J, Demuink E, Zhuang Z, Drinane M, Kauser K, Rubanyi G, et al.Endothelial nitric oxide synthase is critical for ischemic remodeling,mural cell recruitment, and blood flow reserve. Proc Natl Acad Sci2005;102:10999-1004.

20. Smith R, Lin K-F, Agata J, Chao L, Chao J. Human endothelial nitricoxide synthase gene delivery promotes angiogenesis in a rat model ofhindlimb ischemia. Arterioscler Thromb Vasc Biol 2002;22:1279-85.

21. Brevetti LS, Chang DS, Tang GL, Sarkar R, Messina LM. Overexpres-sion of endothelial nitric oxide synthase increases skeletal muscled bloodflow and oxygenation in severe rat hind limb ischema. J Vasc Surg2003;38:820-6.

22. Van Weel V, Toes RE, Seghers L, Deckers MR, de Vries PH, Eilers J, etal. Natural killer cells and CD4� T-cells modulate collateral arterydevelopment. Arterioscler Thromb Vasc Biol 2007;27:2310-8.

23. Alderton WK, Cooper CE, Kowles RG. Nitric oxide synthases: struc-ture, function and inhibition. Biochem J 2001;357:593-615.

24. Heil M, Ziegelhoeffer T, Mees B, Schaper W. A different outlook on therole of bone marrow stem cells in vascular growth: bone marrow deliverssoftware not hardware. Circ Res 2004;94:573-4.

25. Böger R, Bode-Böger S, Thiele W, Junker W, Alexander K, Frölich J.Biochemical evidence of impaired nitric oxide synthesis in patients withperipheral arterial disease. Circulation 1997;95:2068-73.

26. Henry T, Annex B, McKendall G, Azarin M, Lopex J, Giordano F, et al.The VIVA trial: vascular endothelial growth factor in ischemia forvascular angiogenesis. Circulation 2003;107:1359-65.

27. Simons M, Annex B, Laham R, Kleiman N, Henry T, Dauerman H, etal. Pharmacological treatment of coronary artery disease with recombi-nant fibroblast growth factor-2: double-blind, randomized, controlledclinical trial. Circulation 2002;105:788-93.

28. Hiasa KI, Ishibashi M, Ohtani K, Inuoe S, Zhao Q, Kitamoto S, et al.Gene transfer of stromal cell-derived factor-1a enhances ischemic vas-culogenesis and angiogenesis via vascular endothelial growth factor/endothelial nitric oxide synthase-related pathway: next-generation che-mokine therapy for therapeutic neovascularization. Circulation 2004;109:2454-61.

29. Ahmad S, Hewett PW, Wang P, Al-Ani B, Cudmore M, Fujisawa T, etal. Direct evidence for endothelial vascular endothelial growth factorreceptor-1 function in nitric oxide-mediated angiogenesis. Circ Res2006;99:715-22.

Submitted May 20, 2009; accepted Aug 12, 2009.