Main

In full-term infants, closure of the ductus arteriosus occurs in two phases after birth: 1) initial muscular constriction of the vessel wall that causes functional obliteration of the lumen in the first hours after delivery, and 2) extensive neointimal thickening that produces permanent occlusion of the lumen over the next week. Neointimal mound formation begins with the separation of luminal EC from the underlying internal elastic lamina and the accumulation of hyaluronan in the expanded subendothelial space and inner muscle media. Luminal EC, then, detach from the basal lamina and migrate into the subendothelial region, where they are joined by SMC that have migrated from the muscle media(16). Associated with the accumulation of EC and SMC in the subendothelial region are extensive changes in the amounts and types of extracellular matrix (e.g. fibronectins, laminins, and collagens) that surround the cells(1, 2, 7). In premature infants, despite functional obliteration of the lumen by smooth muscle constriction, neointimal mounds frequently fail to develop, resulting in vessel reopening(8).

The migration of EC and SMC during ductus closure requires the presence of cell surface receptors that interact with the surrounding extracellular matrix. Members of the integrin family of transmembrane receptors play an essential role in creating traction with the surrounding matrix and providing signals for cytoskeleton realignment and initiation of cortical flow. The integrins bind to extracellular matrix molecules through their external domain and to cytoskeletal structures intracellularly(9, 10). Integrins are heterodimers composed of two subunits, α andβ. The ligand specificity and signaling ability of integrins are determined by the combination of α and β subunits that make up the heterodimers. Extracellular matrix molecules interact with several different integrins (fibronectins: α3β14β1, α5β18β1, αvβ1vβ3, αvβ5vβ6, α4β7, andαiibβ3; collagens:α1β1, α2β13β1, αvβ3;laminins: α1β12β1, α3β16β1, α7β1vβ3, and α6β4). The exact reason for this redundancy is unclear.

EC, in vivo, have been found to express variable amounts of the following integrin subunits: α1, α23, α5, α6, αv1, β3, β4, and β5(1118). Although there are only a limited number of reports concerning the expression of integrins in EC from normal vascular tissue, a general pattern seems to be emerging: EC in growing capillary beds (as found in the fetus) usually express a larger repertoire of integrins than quiescent EC, such as those found lining the lumen of large vessels or those found in adult capillary beds(1117, 19). A somewhat different set of integrin subunits has been found on vascular SMC,in vivo: α1, α3, α58, α9, αv, β1, andβ3(11, 18, 2023).

The purpose of the present study was to identify the repertoires of integrins expressed by EC and SMC of the ductus arteriosus, in vivo, and to compare them with the pattern expressed by the corresponding cells of a neighboring vessel, the aorta. We especially wanted to identify which, if any, integrins were newly expressed during vessel remodeling, and to see whether changes in ductus structure could occur without simultaneous changes in the expression of cell-specific integrins. We used ductus arteriosus obtained from full-term, preterm, fetal, and newborn rhesus monkeys to examine the effects of gestational age and postnatal age on integrin expression and ductus remodeling. Finally, we compared our findings of integrin expression,in vivo, with previous findings that examined integrin functionin vitro, to gain insight into the relative importance of the different integrin families in mediating cell migration and closure of the ductus arteriosus.

METHODS

Tissue preparation. In this study, we used ductus arteriosus obtained from fetal, preterm newborn, and full-term newborn rhesus(Macaca mulatta) monkeys. Monkeys, obtained from the California Regional Primate Research Center at the University of California at Davis, were used because of the broad panel of antibodies available to study primate integrin receptors. Unfortunately, these reagents are not as extensively available for other species. Animal care, surgery, and necropsy were performed according to standard Primate Center operating procedures.

Full-term newborn monkeys (n = 3) were given an overdose of pentobarbital sodium at 96 h after delivery. Fetal monkeys (n = 4) were delivered by cesarean section, with the dam under general anesthesia, and were given an overdose of pentobarbital sodium before they began breathing air. Preterm newborns (n = 8) were delivered at gestational age 128 d (78% of gestation), intubated with a 2-mm endotracheal tube, and transferred to a radiant warmer in the primate intensive care nursery. Mechanical ventilation, blood transfusions, and fluid infusion rates, in addition to measurements of heart rate, blood pressure, blood gases, electrolytes, and glucose, were performed in a manner similar to that described for human infants(24). Ventilator management was designed to maintain the arterial blood gases in the following ranges: Pao2 = 6.6-13.3 kPa (50-100 torr), Paco2 = 4.7-7.3 kPa (35-55 torr), and pH = 7.25-7.45. Ventilator settings and Fio2 were adjusted according to blood gas measurements. The end-expiratory pressure was maintained at 0.4 kPa(4 cm H2O) and the inspiratory time at 0.35-0.40 s for the duration of the study. At 96 h after delivery, the animals were killed with an overdose of pentobarbital sodium. Ventilator settings just before euthanasia were: peak inspiratory pressure, 27 ± 10 (mean ± SD) cm H2O; rate, 31 ± 18 breaths/min; and Fio2, 0.38 ± 0.19.

After dissection, the ductus arteriosus was rinsed in physiologic saline, embedded in Tissuetek (Miles Inc., Elkhart, IN), and immediately frozen in liquid nitrogen. Specimens were stored at -80°C until analyzed.

Immunohistochemistry. The protocol for immunohistochemical studies of integrins was similar to methods we have reported previously(3). In the current studies, it was necessary to use nonfixed, frozen tissue because of the loss of antigenicity of many of the integrins after fixation(25). Briefly, frozen sections(6 μm) of ductus arteriosus were cut, air-dried, and immersed in acetone at-20°C for 10 min. Endogenous tissue peroxidase was quenched with 0.3% hydrogen peroxide, and nonspecific binding was blocked with normal goat serum(5%). Sections were incubated with primary antibodies diluted in PBS containing 1 mg/mL BSA for 2 h, rinsed with the same solution for 30 min, and incubated with biotinylated secondary antibody (either goat-anti-mouse IgG, goat-anti-rabbit IgG, or goat-anti-rat IgG: 1/200 dilution, Vectastain ABC Elite kit; Vector Laboratory, Burlingame, CA) for 60 min. The sections were then exposed to avidin-biotin-complex (ABC Elite kit; Vector Laboratory) and reacted with diaminobenzidine according to the manufacturer's recommendations. Sections were counterstained with hematoxylin. Control sections were incubated with similar concentrations of either mouse, rabbit, or rat IgG. Sections were photographed using a Leitz Orthoplan 2 microscope. Samples from at least1) two different fetal monkeys, 2) two full-term newborns,3) two preterm newborns with a closed ductus (see below), and4) three preterm newborns with a patent ductus (see below) were processed together in the same histochemical assay for a particular antibody. Assays for any given antibody were reproduced on three separate occasions. Ductal tissue from all of the animals in a given group (see below) showed the same pattern of staining.

Antibodies. Mouse monoclonal anti-human α1 (TS2/7)(26) was provided by Dr. M. Hemler (Dana Farber Cancer Institute, Boston, MA). Mouse monoclonal anti-α2 (VM-1) and anti-α3 (VM-2) was provided by Dr. V. Morhenn (SRI International, Menlo Park, CA)(27). Mouse monoclonal anti-α5 (P1D6) was purchased from Chemicon International (Temecula, CA)(28). Rat monoclonal anti-α6 (G0H3) was provided by Dr. Arnoud Sonnenberg (Central Laboratory of the Netherlands Red Cross Blood Transfusion Service, Amsterdam, The Netherlands)(29). Affinity-purified rabbit polyclonal antibodies to α8(22) and to α9(23), in addition to mouse monoclonal anti-β5 (PIF6)(30) and anti-β6(E7P6G10P0)(30) have been described previously. Mouse monoclonal anti-αv (LM 142) was provided by Dr. David Cheresh (Research Institute of Scripps Clinic, LaJolla, CA)(31). Rat monoclonal anti-β1(AIIB2) was provided by Dr. Caroline Damsky (University of California, San Francisco, CA)(32). Mouse monoclonal anti-β3 was purchased from Immunotech, Inc. (Westbrook, MA). Mouse monoclonal anti-β4 (3E1) was purchased from Telios Pharmaceuticals(San Diego, CA). Mouse monoclonal anti-smooth muscle actin was purchased from Sigma Chemical Co. (St. Louis, MO). Rabbit polyclonal antibodies to von Willebrand factor were purchased from Dakopatts, Denmark. Mouse monoclonal anti-PECAM was purchased from R and D Systems (Minneapolis, MN).

SMC were identified by their reaction with anti-SMC-actin antibody. EC were identified by their reaction with antibodies to two EC markers: von Willebrand factor and PECAM(33). Although data are presented only for PECAM, the staining of consecutive sections with antibodies to von Willebrand factor was identical to the PECAM staining. We have assumed that integrin heterodimers, and not single α or β subunits, are being detected(34, 35).

RESULTS

Morphologic changes in the full-term and preterm ductus arteriosus. The fetal ductus arteriosus of the rhesus monkey at 0.78 gestation had a single layer of EC separated from the smooth muscle media by a continuous internal elastic lamina (Fig. 1,A andD). Occasional SMC penetrated beyond the internal elastic lamina and lay directly beneath the luminal EC. This arrangement of cells did not change as the rhesus monkey fetus approached full-term (data not shown). This failure to develop a neointima, as the rhesus fetus approached term, is in contrast with some other species (e.g. sheep and rabbit), where a modest neointima has been reported to develop during late gestation(6, 36).

Figure 1
figure 1

SMC and EC alterations during ductus arteriosus closure. SMC are stained brown with α-smooth muscle actin antibody(A-C). (A) Fetal ductus (L = lumen).(B) A 4-d-old full-term newborn. SMC have migrated into the neointima (NI), crossing the internal elastic lamina (dashed line). Nonstained EC nuclei fill the central portion of the neointima that obliterates the ductus lumen. (C) A 4-d-old preterm newborn ductus with a closed lumen. EC are stained brown with PECAM antibody (D-J).(D) Fetal ductus. (E) A 4-d-old full-term newborn. EC, which fill the central portion of the neointima, migrate into the subendothelial space, across the internal elastic lamina (dashed line), and into the inner muscle media (arrows). (F) A 4-d-old preterm newborn with a closed lumen. Invasion of the muscle media by vasa vasorum (v.v.) during ductus closure (G-J). Each figure (G-J) is photographed at the same magnification.(G) Fetal ductus. Vasa vasorum do not penetrate muscle media(M) and are found only in the adventitia (A). (H and I) Partially closed premature ductus. Vasa vasorum begin to invade the outer muscle media. A single layer of EC continues to line the ductus lumen. (J) Obliterated ductus lumen in a premature ductus. Vasa vasorum invade the muscle media and luminal EC invade the neointima and muscle media (dashed line = internal elastic lamina).(Horizontal bar represents μm).

By 4 d after delivery, the ductus arteriosus of the full-term newborn was markedly constricted and had no lumen by gross visual examination. SMC had migrated from the tunica media to the subendothelial space, forming a multicell layer thick neointima (Fig. 1B) that helped to obliterate the ductus lumen. Over the same period of time, luminal EC migrated into the subendothelial space, joining the invading SMC as part of the neointima (Fig. 1E), and continued to migrate through the internal elastic lamina to invade the inner one-third of the muscle media(Fig. 1E).

The ductus arteriosus of all eight preterm newborns appeared to be closed, by clinical examination at the time of necropsy (4 d). In six newborns, clinical evidence of ductus arteriosus patency (murmur and widened pulse pressure) disappeared by 24 h after delivery; in two other preterm newborns, clinical signs persisted for 72 h before disappearing. However, on visual examination at the time of necropsy, only three of the eight preterm monkeys had actually closed their ductus lumens. The other five still had perceptible lumens, even though the vessels were markedly to modestly constricted. The histologic appearance of the three preterm ductus arteriosus with closed lumens was similar to what we saw in the 4-d-old full-term ductus. There was neointimal formation composed of both migrating SMC (Fig. 1C) and EC (Fig. 1F). In contrast, the five preterm ductus that were constricted but still patent failed to develop a neointima(see below).

In the fetus, a single row of vasa vasorum traversed through the adventitial compartment at the outer edge of the muscle media(Fig. 1G). The thickness of the fetal ductus muscle media was similar to that of the aorta. After delivery, the muscle media of the ductus became markedly thickened due to new matrix production and a combination of circumferential constriction and shortening of the vessel(1, 7, 8). As the vessel's lumen closed and the thickness of the muscle media increased, the vasa vasorum began to invade the wall of the muscle media (Fig. 1,H,I,andJ). Ultimately, the entire outer muscle media became hyperemic(Fig. 1J)(3, 8, 37). If the lumen of the ductus remained patent, then the vasa vasorum failed to invade the muscle media (Fig. 1,H andI).

Expression of integrins in EC of the fetus and newborn. EC that lined the lumen of the fetal ductus arteriosus had weak expression of theβ1 integrin α1, and minimal to negligible expression of the other members of the β1 family(α2v) (Fig. 2; see alsoFigs. 3,A andG,4A,and5A). Consistent with the low level of αv staining (Fig. 2; see also Fig. 4A), there was minimal to negligible staining of the αv integrins: αvβ3vβ5, and αvβ6(Fig. 2; see also Figs. 4C and5G). There was also negligible staining of theα6β4 integrin (Fig. 2; see alsoFig. 4B). The EC that lined the lumen of the fetal aorta had the identical integrin staining pattern as the ductus luminal EC (data not shown).

Figure 2
figure 2

Distribution of integrin subunits in EC and SMC of the fetal, preterm newborn, and full-term newborn ductus arteriosus. Luminal and vasa vasorum (capillary) EC, in addition to SMC of the muscle media and neointima, were scored for staining: - = no staining; = weak staining;□ = moderate staining; • = intense staining. Capillary EC integrin expression was the same in both the fetal and the newborn ductus.

Figure 3
figure 3

Luminal EC of the ductus arteriosus increase their expression (brown stain) of the integrin α2β1(A-F) and α6 subunit (G-L) during ductus closure. α2β1: (A) In fetus, luminal EC have minimal α2β1 staining (L = lumen).(B) In full-term newborn, EC of the neointima (NI) and those migrating across the internal elastic lamina (dashed line) into the muscle media (arrows) stain intensely forα2β1. (C) In fetus, capillary EC of adventitial vasa vasorum (v.v.) stain intensely forα2β1. (D) In preterm newborn with patent ductus, luminal EC have minimal α2β1 staining.(E) In preterm newborn with closed ductus, EC of the neointima and those migrating into the muscle media (arrows) stain intensely forα2β1. (F) In full-term newborn, EC of the vasa vasorum (v.v.) and their capillary extensions in the adventitia and muscle media stain for α2β1. α6:(G) In fetus, luminal EC have minimal α6 staining.(H) In full-term newborn. EC of the neointima and those migrating across the internal elastic lamina into the muscle media (arrows) stain intensely for α6. (I) In fetus, luminal EC(lum. ec) have minimal α6 staining, whereas EC of the vasa vasorum (v.v.) stain intensely (A = adventitia;M = muscle media). (J) In preterm newborn with patent ductus, luminal EC have minimal α6 staining. (K) In preterm newborn with closed ductus, EC of the neointima and those migrating into the muscle media (arrows) stain intensely for α6.(L) In full-term newborn, vasa vasorum (v.v.) invading the outer muscle media stain intensely for α6. (Horizontal bar represents μm.)

Figure 4
figure 4

Luminal EC of the ductus arteriosus increase their expression of α1, α3, αv4, and β5 during ductus closure. In the fetus, EC lining the ductus lumen have minimal staining for the integrin subunitsαv (A), β4 (B), andβ5 (C) (L = lumen). In contrast, capillary EC of the adventitial vasa vasorum (v.v.) stain intensely forαv (D), β4 (E), andβ5 (F). In the full-term newborn, after ductus closure, neointimal (NI) EC stain intensely for αv(G), β4 (H), and β5 (I). In addition, neointimal EC stain intensely for α3 (J) and α1 (K), but do not stain for α8(L). SMC of the fetus and full-term newborn stain for [Illegible Text] (A and G), β5 (C andI), α3 (J), [Illegible Text] (K), and α8 (L). (Horizontal bar representsμm.)

Figure 5
figure 5

SMC of the ductus arteriosus increase their expression(brown stain) of the integrins α5β1 (A-F) and αvβ3 (G-L) during closure.α5β1: (A) In fetus, ductus smooth muscle and luminal EC have minimal α5β1 staining(L = lumen). (B) In full-term newborn, SMC in the muscle media (M) and neointima (NI) stain intensely forα5β1. Arrowheads point to minimally stained EC in the central region of the neointima. Dashed line outlines internal elastic lamina. (C) In full-term newborn, SMC of the outer muscle media (OM) and of the vasa vasorum (v-sm) in the adventitia (A) stain intensely for α5β1.(D) Im preterm newborn with patent ductus, smooth muscle and EC have minimal α5β1 staining. (E) In preterm newborn with closed ductus, changes are similar to those in B(full-term newborn). (F) In full-term newborn, smooth muscle in the neointima and throughout the muscle media stain forα5β1, αvβ3: (G) In fetus, ductus smooth muscle and luminal EC have minimalαvβ3 staining. (H) In full-term newborn, SMC in the muscle media and neointima stain intensely forαvβ3. EC in the central region of the neointima(arrowheads) are not stained. (I) In full-term newborn, SMC of vasa vasorum (v-sm) stain for αvβ3.(J) In preterm newborn with patent ductus, SMC and EC have minimalαvβ3 staining. (K) In preterm newborn with closed ductus, SMC in the neointima and inner half of the muscle media stain intensely for αvβ3. (L) In full-term newborn ductus and aorta, SMC in the neointima and inner half of the ductus muscle media (Duct M) stain intensely forαvβ3. In contrast, SMC of the aortic muscle media do not stain. (Ao L = aortic lumen; horizontal bar representsμm.)

The weak expression of integrin subunits on the luminal EC was in sharp contrast to their strong expression in the capillaries found in the adventitia of both the fetal ductus and aorta (Fig. 2). Capillary extensions of the vasa vasorum strongly expressed several β1 integrins: α1, α2, α3, α6, and αv (Fig. 2; see alsoFig. 3,C,I,and4D). They also expressed theαv integrin, αvβ5 (but had only negligible expression of the αvβ3 integrin(Fig. 2; see also Fig. 4F). Capillary EC stained strongly for the α6β4 integrin(Fig. 2; see also Fig. 4E).

After delivery, there was no change in the pattern of integrin expression in the EC that lined the lumen of the aorta (data not shown). In contrast the luminal EC of the full-term ductus stained strongly for the integrins:α1β1, α2β13β1, α6β16β4, and αvβ5. This pattern of integrin expression was identical to that found on the invading capillaries of the vasa vasorum (Fig. 2; see alsoFigs. 3,B andH,and4,G,H,I,J,andK).

Expression of integrins in vascular SMC of the fetus and newborn. SMC of the fetal ductus arteriosus muscle media had marked positive staining for several integrins of the β1 family:α1, α3, α8, and α9(Fig. 2). They also expressed the αv integrin, αvβ5 (Fig. 2; see alsoFig. 4,A andC). The αv integrin,αvβ1, also may have been present; however, this could not be demonstrated because specific antibodies toαvβ1 are not available. Previous studies have used immunoprecipitation to document the presence of αvβ1 in cultured ductus arteriosus SMC(38). The repertoire of integrins expressed by SMC of the fetal aorta was identical to that of fetal ductus arteriosus (data not shown).

After delivery, there was no change in the pattern of integrin expression in SMC of the full-term aorta (data not shown). In contrast, SMC of the full-term ductus increased their repertoire of expressed integrins. Throughout the muscle media, SMC of the remodeling ductus became positive for the fibronectin receptor, α5β1 (Fig. 2; see also Fig. 5,B andF). In addition, in association with ductus closure, SMC of the vasa vasorum also became positive for α5β1 (Fig. 5C).

SMC of the closing ductus markedly increased their expression of the promiscuous integrin receptor αvβ3(Fig. 2; see also Fig. 5,H andL). Localization of the αvβ3 receptor was strongest on SMC of the inner half of the closing ductus (Fig. 5L). In addition, in association with ductus closure, SMC of the vasa vasorum also became positive for αvβ3 (Fig. 5I).

Expression of integrins in the preterm ductus arteriosus. Five of the preterm neonatal ductus failed to obliterate their lumen by muscle constriction and failed to develop a neointima. At 4 d after birth, the luminal EC and the SMC of these partially patent vessels still had the same pattern of integrin expression as the fetal ductus. Both cell types failed to undergo any of the changes in integrin expression found in the full-term ductus at the same postnatal age. The luminal EC did not increase their expression of α2 or α6 (Fig. 3,D andJ) or any of the other integrin subunits found in the full-term, closing ductus arteriosus (Fig. 2). Similarly, the SMC did not increase their expression of either α5β1(Fig. 5D) or αvβ3(Fig. 5J).

Three of the preterm ductus had lumens that were completely obliterated by muscle constriction and neointimal formation. The repertoire of integrins expressed by the luminal EC (Fig. 2; see alsoFig. 3,E andK) and the SMC (Fig. 2; see also Fig. 5,E andK) in these three preterm vessels was identical to that expressed by the cells of the full-term, closed ductus arteriosus.

DISCUSSION

After ductus closure, luminal EC switch their integrin phenotype from a quiescent pattern to one found on growing capillary EC. In the present study we found no differences during the last 25% of gestation, in the repertoire of integrins expressed by either the fetal ductus or the fetal aorta. EC lining the lumen of the fetal ductus and aorta had only a limited repertoire of integrins (small amounts of α1β1 and negligible amounts of the other integrins) (Fig. 2). This pattern was similar to that observed in quiescent adult EC(14, 16, 17).

In contrast, the growing capillary extensions of the vasa vasorum expressed a broad range of integrins, in both the fetal ductus arteriosus and the fetal aorta (α1β1, α2β13β1, α6β1vβ1, α6β4, andαvβ5). A similar pattern has been described in capillaries of other fetal organs(12, 13).

After delivery, in association with closure of the ductus lumen, there were extensive changes in the expression of integrins by EC that lined the ductus lumen. Concurrent with their invasion into the neointimal mounds, the luminal EC changed their phenotype and expressed the full repertoire of integrins found on growing capillary EC (Fig. 2). The factors that initiated these changes in integrin expression are currently unknown. Although EC from large vessels have a more restricted integrin repertoire than those of growing capillaries, they can be induced to express capillary-specific integrins when exposed to angiogenic growth factors(14). After delivery, ductus constriction produces loss of luminal blood flow and hypoxia of the inner vessel wall; this, in turn, coincides with the production of growth factors (such as TGFβ)(3) and vascular endothelial growth factor (our unpublished observations), which may be responsible for initiating or directing these changes.

Recently, Brooks et al.(15) demonstrated that the αvβ3 integrin appeared to be essential for angiogenesis and new vessel growth. Although we observed an increase inαvβ3 expression in SMC of the closing ductus, we found only negligible expression on EC of either the growing capillaries of the vasa vasorum or the migrating luminal EC (Figs. 2 and5H). Although there was an increase in αv integrins in the migrating EC of the ductus arteriosus (Fig. 4G), this appeared to be mainly due to an increase in αvβ5(Fig. 4I). The αv integrins play a role in intracellular calcium signaling. This can be mediated by eitherαvβ3 or other αv integrins(39). Whereas β1 integrins appear to be more involved with cell adhesion, both αvβ3 andαvβ5 integrins contribute to cell migration(40, 41). It is possible that EC of the closing ductus arteriosus use αvβ5 instead ofαvβ3 to facilitate their migration into the neointima and muscle media. Our findings do not rule out the possibility thatαvβ3 could have been transiently expressed, at an earlier time point after delivery, or that it functions in very low amounts in the ductus arteriosus.

After ductus closure, SMC acquire new, promigratory integrins. SMC of the fetal ductus arteriosus express the same repertoire of integrins expressed by those of the fetal aorta (α1β13β1, α8β19β1, and the αv integrinsαvβ5 and possibly αvβ1). This pattern is nearly identical to what has previously been reported for human vascular SMC in vivo(11, 18, 2023).

After birth, SMC in the closing ductus arteriosus change their phenotype so that they express two additional integrins: the classical receptor for fibronectin, α5β1, and the promiscuous integrin,αvβ3, which recognizes a variety of RGD-containing matrix proteins (Fig. 5,B andH). Fibronectin and itsα5β1 receptor play an important role in facilitating ductus SMC migration in vitro(42, 43). The up-regulation of α5β1 in ductus SMCs may not be unique to SMC; α5β1 appears to be up-regulated in migrating epithelial cells during wound healing(4446) and is present in granulation tissue(19).

SMC of the ductus neointima and inner half of the muscle media also increase their expression of αvβ3(Fig. 5H). We have previously shown that migration of isolated ductus SMC depends to a large extent on theαvβ3 receptor, even though it plays only a minor role in SMC adhesion(38). The exact mechanism through which this occurs is still unclear. Recent studies have found that theαvβ3 integrin can activate intracellular calcium signaling pathways(39, 47) and regulate the functions of the α5β1 fibronectin receptor(48). Therefore, when SMC are induced to migrate into the expanding subendothelial space of the closing ductus arteriosus, they begin to express the same integrins that are functionally necessary for them to migratein vitro.

Preterm monkeys that fail to close their ductus lumen do not alter their pattern of integrin expression. In premature infants, neointimal mounds frequently fail to develop despite apparent muscular constriction of the ductus after birth. This may be due to an inability of the premature ductus to obliterate luminal blood flow completely, because even after clinical signs have disappeared, 23% of human preterm infants continue to have an asymptomatic, persistently patent ductus lumen(49). Whereas most infants without evidence of luminal patency will remodel their ductus, only 50% of those with a persistently patent lumen will do so(49). Our present observations, in preterm monkeys, are consistent with these findings. In the preterm monkey, changes in SMC and EC integrin profile and neointima formation occurred only in those ductus in which the lumen appeared closed. Associated with the obliteration of the lumen, active angiogenesis also developed in the outer vessel wall(Fig. 1J). In contrast, despite significant muscular constriction, ductus with a persistently patent lumen still displayed a fetal pattern of integrin expression and failed to develop neointimal mounds(Fig. 2; see also Figs. 3,D andJ,and5,D andJ). These vessels also failed to develop active angiogenesis, presumably because their inner muscle wall did not become sufficiently hypoxemic and still was able to receive oxygen and nutrients through the persistently patent lumen (Fig. 1,H andI).

In conclusion, our findings show that EC and SMC change from a quiescent phenotype to an activated state to produce the intimal thickening required for ductus closure. Both EC and SMC increase their integrin repertoire during ductus closure to mimic the pattern seen in migrating cells. These changes appear at the same time EC and SMC invade their neighboring compartments. No evidence of intimal thickening occurs in the absence of changes in integrin expression. Preterm newborns, with a persistently patent ductus lumen, fail to develop these changes in integrin expression; this finding suggests that the changes in integrin expression are induced by hypoxic ischemia of the vessel wall produced during obliteration of the ductus lumen. Identification of the cytokines and growth factors released within the hypoxic wall of the ductus should help us understand the changes in integrin expression. It may also help us understand the regulation of angiogenesis and SMC motility in other tissue sites after injury.