Download Origin, Differentiation, and Maturation of Human Pulmonary Veins

Origin, Differentiation, and Maturation of Human Pulmonary Veins
Susan M. Hall, Alison A. Hislop, and Sheila G. Haworth
Unit of Vascular Biology and Pharmacology, Institute of Child Health, University College, London, United Kingdom
Recent studies on human embryonic and fetal lungs show that
the pulmonary arteries form by vasculogenesis. Little is known
of the early development of the pulmonary veins. Using immunohistochemical techniques and serial reconstruction, we
studied 18 fetal and neonatal lungs. Sections were stained with
antibodies specific for endothelium (CD31, von Willebrand factor) and smooth muscle ( and smooth muscle actin,
smooth muscle myosin, calponin, caldesmon, and desmin) and
antibodies specific for the matrix glycoprotein tenascin, the
receptor protein tyrosine kinase EphB4, and its ligand ephrinB2.
Kiel University–raised antibody number 67 (Ki67) expression
allowed qualitative assessment of cell replication. By 34 d gestation, there was continuity between the aortic sac, pulmonary arteries, capillaries, pulmonary veins, and atrium. The pulmonary veins formed by vasculogenesis in the mesenchyme
surrounding the terminal buds during the pseudoglandular
period and probably by angiogenesis in the canalicular and alveolar stages. EphB4 and ephrinB2 did not distinguish between presumptive venous and arterial endothelium as they
do in mouse. All venous smooth muscle cells derived directly
from the mesenchyme, gradually acquiring smooth muscle specific proteins from 56 d gestation. Thus, both pulmonary arteries and veins arise by vasculogenesis, but the origins of their
smooth muscle cells and their cytoskeletal protein content are
different.
Little is known about the embryonic and fetal development of human pulmonary veins. Historically, the pulmonary veins have received less attention than the pulmonary
arteries, being regarded as passive conduits, whereas the
arteries regulate pulmonary vascular resistance. However,
recent studies have shown that despite being thin walled,
isolated pulmonary veins show a greater response to both
contractile and relaxant agonist stimulation and greater
release of nitric oxide than pulmonary arteries (1). It is
now apparent that the pulmonary veins form an active segment of the pulmonary circulation.
In a recent study of human embryos, serial reconstruction showed that the pulmonary arteries form by vasculogenesis from the spanchnopleural mesenchyme (2). Distinguishing presumptive arterial and venous endothelium from
each other in the mesenchyme can be problematic. However,
in mouse embryos, endothelial cells destined to become
systemic venous cells expressed the tyrosine kinase receptor EphB4, whereas its cognate ligand, ephrinB2, was expressed on predestined systemic arterial endothelial tubes
(3–5). Thus, in mice, venous and arterial endothelial cells
(Received in original form October 22, 2001)
Address correspondence to: Dr. S. M. Hall, Unit of Vascular Biology and
Pharmacology, Institute of Child Health, 30 Guilford St., London WC1N
1EH, UK. E-mail: [email protected]
Abbreviations: alpha smooth muscle specific actin, -SM actin; cluster of
differentiation number CD31, CD31; gamma smooth muscle specific actin, -SM actin; Kiel University-raised antibody number 67, Ki67; smoothmuscle–specific myosin heavy chain SM1, SM-1; vascular endothelial
growth factor, VEGF.
Am. J. Respir. Cell Mol. Biol. Vol. 26, pp. 333–340, 2002
Internet address: www.atsjournals.org
are molecularly distinct from the earliest stages of development (5). The Eph receptor tyrosine kinase family and
their ephrin ligands have been shown in early vertebrate
development to be responsible for many boundary and
guidance processes (6). In particular, EphB4 and ephrinB2
are essential for the development of the cardiovascular
system in mouse embryos. During the development of pulmonary arteries, the airways appear to act as a template.
However, in the mature lung, the veins lie at some distance
from airway and artery, in the connective tissue septa, suggesting that the airways may have less of a direct influence
on the control of vein development in utero.
In the present study we hypothesized that the endothelial expression of EphB4 and ephrinB2 in the lung mesenchyme of the human embryo would distinguish between
presumptive pulmonary venous and pulmonary arterial
endothelium. Furthermore, we considered that the difference in spatial relationship between veins with airways
and arteries with airways would be reflected in the origin
and composition of the venous smooth muscle. We used
immunohistochemical techniques to study serial sections
of human embryonic and fetal lungs. Reconstruction of
the serial sections showed that in human embryos the pulmonary veins, like the pulmonary arteries, are originally
derived by vasculogenesis from the mesenchyme surrounding the terminal airways. Reconstruction also confirmed
the presence of the postnatal anatomic relationship of arteries, veins, and capillaries from embryonic life. It was not
possible to discriminate presumptive veins from arteries
by their expression of EphB4 and ephrinB2.
Materials and Methods
Human fetuses aged 34 (n 2), 38 (n 2), 44 (n 1), 47 (n 1),
56 (n 1), 84 (n 4), and 98 (n 1) d were obtained from the
Medical Research Council Human Embryo Collection (London,
UK). The fetuses were staged by limb bud and facial features (7).
In addition, lungs from three fetuses aged 105, 119, and 140 d
were obtained from the Anatomy Department of the University
of Leiden (Leiden, the Netherlands), and their collection and use
was approved by the Medical Ethics Committee of the Leiden
University Medical Centre. The whole fetus or lung of all but two
84-d-old fetuses were fixed in 4% paraformaldehyde for 24 h at
4C and then rinsed in several changes of 70% alcohol before being processed for wax histology. In addition, we examined postmortem lung biopsies taken from three neonates who died with
normal lungs. This tissue was taken within 24 h of death and fixed
in 10% buffered formaldehyde for a minimum of 24 h before processing for routine histology. Serial 5- m wax sections were cut
through the thoracic region of each fetus or lung tissue sample.
The sections were used for immunohistochemical staining
with antibodies specific for either vascular endothelial or smooth
muscle cells (2), for the receptor EphB4 and its ligand ephrinB2,
and for tenascin, a matrix glycoprotein associated with angiogenesis (8, 9) (Table 1). In addition, tissue sections were stained with
hematoxylin and eosin and with Miller’s elastic van Gieson stain.
Sections were also stained for replicating cells by expression of
334
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 26 2002
Kiel University-raised antibody number 67 (Ki67), which labels all
cells not in the Go phase of mitosis (Table 1). Before incubation
with the primary antibody, the sections were first dewaxed, and
antigen unmasking was performed by autoclaving at 121 C for 14
min in citric acid buffer (2.1 g citric acid in 1 liter distilled water,
pH adjusted to 6.0). Endogenous peroxidase was then blocked by
incubation in 0.3% hydrogen peroxide in methanol and nonspecific binding blocked by incubation with serum-free protein block
(Dako UK, Cambridgeshire, UK). Sections were incubated in
100 l of primary antibody diluted in phosphate buffered saline
with 0.02% bovine serum albumin for 1 h at room temperature.
Antibody binding was visualized with diaminobenzidine using a
biotin-conjugated relevant secondary antibody and avidin streptavidin amplification (StrepABComplex, Dako UK). Sections were
lightly counterstained with hematoxylin and eosin. For each gestational age group, at least eight sections were studied with each
antibody.
The lungs of two 84-d-old fetuses were mounted on cork discs
in Cryo-M-Bed (Bright Instrument, Cambridge, UK) and snap
frozen in isopentane cooled in liquid nitrogen. Serial 10- m sections of lung tissue were cut in a transverse plane, air dried, and
fixed briefly with paraformaldehyde and stained for ephrinB2
(Table 1).
In addition to conventional microscopic study, three-dimensional reconstructions were made from images acquired using a
Zeiss Axioskop microscope, 10 objective, Hamamatsu camera,
and the OpenLab Program (Improvision, Warwick, UK).
Results
Origin and Distribution of Pulmonary Veins
Reconstruction of serial sections showed that at 34 d of
gestation the early lung bud was entirely enclosed within
the splanchnopleural mesenchyme, which surrounds the foregut (Figures 1a and 1b). The trachea lay ventral to the
esophagus and gave rise to two bronchi, which curved dorsally to lie on either side of the esophagus. The heart lay
ventral to the lung bud, and at the level of the atria the
dorsal mesocardium was continuous with the splanchnopleural mesenchyme. CD31-labeled endothelial tubes in
the mesenchyme around each lung bud merged on the ven-
tral side of the lung bud and joined to form the pulmonary
venous confluence, which opened into the prospective left
atrial cavity (Figures 1a and 1b). Endothelial tubes also
merged on the lateral side of each bronchus to form a vessel that was in continuity with paired pulmonary arteries
connecting with the aortic sac at the cephalic end of the
trachea. Erythrocytes were present in all vessels, suggesting that blood could circulate through the lung bud from at
least 34 d gestation.
From 38–56 d gestation, the lung buds gradually separated from the oesophageal mesenchyme, and distally in
each lung bud the mesenchyme expanded progressively
around an increasing number of airway generations. This
resulted in a relative narrowing of the hilar region. The extrapulmonary veins lengthened with age as the heart (left
atrium) differentiated and separated from the base of the
lung. Within the mesenchyme surrounding the terminal
airway buds (Figures 1c and 1d), a plexus of CD31-labeled
endothelial tubes or capillaries could be identified. Cell
replication, as assessed qualitatively by the number of cells
reacting positively to Ki67, was greatest in the undifferentiated peripheral mesenchyme surrounding these capillaries (Figure 1e). Beyond the capillary bed, the endothelial
tubes merged to form venules that were separated from
the artery by the airway bud. Serial reconstruction showed
that these venules and capillaries were continuous proximally with the main pulmonary veins and distally with capillaries that were in continuity with pulmonary arteries
(Figure 1c). By 84 d of gestation, veins ran midway between airways where the mesenchymal cell density was
low (Figure 1f). At all ages between 38 and 98 d gestation,
during the pseudoglandular phase of development (10),
continuous coalescence of capillaries at the periphery
lengthened the veins and increased the number of pulmonary vein generations. Thus, the veins appeared to be derived by vasculogenesis.
There was a marked change in the appearance of the
lung after 84 d gestation, when the lung periphery ap-
TABLE 1
Immunohistochemical markers
Antibody
Endothelial specific protein
Monoclonal anti-human CD31
Rabbit anti-human von Willebrand factor
Rabbit polyclonal anti-ephrinB2 human
Goat polyclonal anti-EphB4 human
Smooth muscle specific proteins
Monoclonal anti--SM actin
Monoclonal anti--SM actin
Monoclonal anti-MHC SM-1
Monoclonal anti-calponin
Monoclonal anti-caldesmon
Monoclonal anti-desmin
Matrix protein
Monoclonal anti-human tenascin
Cell replication marker
Monoclonal anti-Ki67
Clone
Dilution
Source
Dako (Ely, Cambs, UK)
Dako (Ely)
Santa Cruz Biotechnology (Santa Cruz, CA)
C16
Prediluted
1:200
1:1,000 (wax)
1:100 (frozen)
1:100
1A4
B4
MCA MH-01
hCP
hHCD
ZC18
1:3,000
1:100
1:1,000
1:1,000
1:400
Prediluted
Sigma (Poole, Dorset, UK)
Dr. J. Lessard (University of Cincinnati, OH)
Yamasa (Tokyo, Japan)
Sigma
Sigma
Zymed Labs (San Francisco, CA)
BC-24
1:500
Sigma
7B11
Prediluted
Zymed Labs
JC/70A
A 0082
P20
Santa Cruz Biotechnology
Hall, Hislop, and Haworth: Fetal Development of Human Pulmonary Veins
335
Figure 1. (a) Drawing derived from a portion of a serial reconstruction of transverse
sections through the thorax of a 34-d-old
embryo showing paired bronchi (black outlines) on either side of the esophagus (oe).
The paired bronchi are surrounded by a
plexus of endothelial tubes (blue), which
coalesce in the dorsal mesocardium (DM)
to form pulmonary veins (pv) that drain
into the left atrium (LA). The presumptive
pleural cavity (pc) surrounds the lung buds.
(b) Drawing derived from a complete serial
reconstruction of the lung bud of the 34-dembryo showing the relationship between
developing pulmonary arterial circulation
(red) at the cephalic end of the lung bud
and the venous circulation (blue) bridging
the dorsal mesocardium to the atrium at the
caudal end of the lung bud. (c) Drawing derived from serial sections of a 56-d-old fetus
that had been labeled for CD31. This illustrates the continuity of a vein (navy), primary capillary plexus (blue), around a terminal airway bud (TB), and an artery (red)
alongside the bronchial smooth muscle of
the peripheral airway (lime green). The
vein and artery were designated as such by
their continuity with more proximal vessels.
The primary capillary bed was made up of
vessels with only an endothelial layer and
no surrounding cells. (d) Photomicrograph
of a longitudinal section through the bronchus at 44 d gestation stained for CD31
(brown) showing the continuity between
the pulmonary vein (pv) at the medial surface of the lung bud (M), the capillary plexus
(cp), and pulmonary artery (pa) at the lateral surface of the lung bud (L). Bar 100 m. Inset, higher magnification of an
adjacent section showing -SM actin positive cells (brown) around the pv. Bar: 25 m. (e) Section through a terminal bud at 56 d gestation stained for Ki67 (brown nuclei are positive for dividing cells) showing a halo of replicating mesenchymal cells around the capillary
plexus (arrowheads) and terminal bud (TB). Bar: 50 m. (f ) Section through a midlung region at 84 d gestation stained for CD31 showing veins (pv) in regions of low cell density midway between airways (br), whereas arteries (pa) lie close to the airways. Bar: 100 m. (g)
Low magnification of lung at 119 d gestation stained for CD31 (brown) showing the lobulated appearance. Veins (pv) are situated
around the edge of the lobules partially surrounded by lymphatic channels (*). Bar: 100 m. (h) Lung at 119 d gestation stained for
CD31 showing a peripheral vein (pv) closely associated with a lymphatic channel (*). A small vein bridges the lymphatic channel and is
continuous with the capillary plexus (arrowheads) around a terminal airway bud (TB). Bar: 50 m. (i) Lung at 119 d gestation stained
for Ki67 (brown nuclei are positive for dividing cells) showing replicating endothelial cells in capillaries close to a terminal bud (TB).
There are few replicating cells in either the surrounding mesenchyme or the pulmonary vein (pv) into which the capillaries drain. (*)
Lymphatic channel. Bar: 50 m. (j) Lung of a neonate stained for CD31. High magnification (inset) of the alveolar region shows the
junction between the alveolar capillary bed and a pulmonary vein (pv). Low magnification shows the connection between such small
pulmonary veins and a larger preacinar vein (pv) lying at some distance from the bronchus (br). Bar: 100 m.
peared to be divided into lobules. In the areas of low cell
density between the airways, there were large clear spaces
lined by CD31-positive cells. These were presumptive lymphatic channels (Figure 1g) forming the prospective septa.
A small amount of collagen was visible in the septa by 105 d.
The lymphatic channels formed an extensive but incomplete
sheath around the larger and more proximal pulmonary
veins, isolating them partially from the surrounding mesenchyme and airways, and were larger than the lymphatic
channels seen postnatally. The lymphatic sheath ended approximately two generations of vein proximal to the capil-
lary plexus. Small veins and venules draining the terminal
airways bridged the sheath and joined the more proximal
veins (Figure 1h).
At 119 d of gestation (canalicular phase [10]), capillaries close to the cuboidal epithelium of terminal buds were
continuous with the venules and then veins. The Ki67 labeling showed that at this age the endothelial cells of these
capillaries had a high replication rate, whereas the surrounding undifferentiated mesenchyme was more sparse
and now had few replicating cells (Figure 1i). This appearance was retained until at least 140 d gestation.
336
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 26 2002
In the neonatal lungs, the capillaries from the alveolar
region drained into venules 5 m in diameter, which in
turn opened into veins 16–20 m in diameter (Figure 1j,
inset). As noted earlier, small veins ran directly from the
alveolar walls and connected with the larger veins lying
within the connective tissue septa (Figure 1j). The lymphatic sheaths around the peripheral veins were still present
two generations from the capillary bed, now within the alveolar region of the lung.
Spatial Relationship Between Tenascin Expression and
Peripheral Vessels
Tenascin is a matrix protein that is expressed in regions of
both cell migration and angiogenesis (11). From 38–98 d
gestation, tenascin was expressed in the mesenchyme around
all airways except the terminal buds (Figure 2a) and was
detectable in the subendothelium of proximal, muscularized veins but not in the arteries. By 105–140 d gestation, it
was found also at the base of the epithelium of the terminal buds (Figure 2b). This distribution was still seen in the
newborn lungs where tenascin was strongly expressed on
the subendothelium of the capillaries and intraacinar veins
(Figure 2c) and was now present in the media of preacinar
veins. Throughout fetal life and at birth, tenascin was not
detectable in the subendothelium or media of any precapillary pulmonary artery (Figure 2c).
Origin and Maturation of Venous Smooth Muscle
A distinct layer of smooth muscle cells, distinguished by
expression of SM actin, outside the endothelium was seen
only from 47 d of gestation and at this time only in the extrapulmonary vein. These smooth muscle cells were flattened and closely applied to the endothelium. This single
layer later became surrounded by a loose meshwork of
elongated but poorly oriented mesenchymal cells that also
stained for -SM actin, presumptive smooth muscle cells
(Figure 3a). By 56 d, a single layer of flattened smooth
muscle cells was found in the intrapulmonary veins. With
age and increase in vein size, the number of muscle layers
increased progressively, and there were six layers in the hilar veins at 98 d of gestation, whereas the more peripheral
veins, lying 2–3 generations postcapillary, had a single
layer of smooth muscle cells. All these smooth muscle cells
appeared to derive from the surrounding mesenchyme.
From 44–140 d of gestation, some endothelial cells of
the smallest venules expressed -SM actin (insets in Figures 1d and 3b) and from 84 d -SM actin positive endothelial cells were seen in the capillaries. By birth, no endothelial cells stained with -SM actin.
Smooth muscle cells of the subendothelial layer gradually acquired other proteins in the order characteristic of
maturating smooth muscle cells (12). First to appear was
the smooth-muscle–specific myosin heavy chain isoform
SM-1 (SM-1), followed by -SM actin (Figure 3) and the
actomyosin regulatory protein calponin. The outermost
layers acquired these proteins slightly later than the inner
layer (Table 2). Caldesmon, another actomyosin regulatory protein, was not expressed in any pulmonary vein, although it was present in the pulmonary arteries at this
time. At 98–140 d, SM-1 immunostaining was no longer
present in proximal veins 70–100 m in diameter (Figure 3d), although -SM actin, -SM actin, and calponin were
still present. The transient disappearance of SM-1 from
the proximal veins was associated with compression of the
smooth muscle cell layers and an increase in lumen size. In
the neonatal lung, all the venous smooth muscle cells again
expressed SM-1 in addition to - and -SM actin. Calponin
was present in some cells in all generations of veins, but
caldesmon was still absent. Thus, all venous smooth muscle was derived directly from the mesenchyme. No -SM
actin positive cells were seen migrating from the airway
smooth muscle to the veins, as has been described for the
arteries (2). This may be as a result of the greater distance
between the veins and airways. Maturation of the smooth
muscle was different in veins and arteries.
In addition to vascular smooth muscle cells, myocardial
muscle was found in the extrapulmonary veins, identified by
the presence of desmin, an intermediate filament protein
expressed only in mature muscle cells and not found in the
pulmonary vasculature during fetal life. Desmin was detected in the atrial myocardium from 38 d gestation, and
discreet dense bundles of desmin-positive cells were present
in the outer part of the wall of the extrapulmonary veins
continuous with the left atrial myocardium from 56 d of
gestation onward (Figure 3e). Staining for -SM actin on
Figure 2. Photomicrographs of sections through
peripheral airways at (a) 98 d gestation, (b) 119 d
gestation, and (c) at birth, stained for tenascin.
Sections were not counterstained. All pictures at
the same magnification. Bar: 100 m. (a) Tenascin
is strongly expressed around the bronchial smooth
muscle and adjacent mesenchyme of proximal airways (br). Staining is weaker peripherally, and
none is detectable around terminal buds (TB) or
the generation proximal to them. pv, pulmonary
vein. (b) Tenascin is now also strongly expressed
at the base of the bronchial epithelium of the terminal buds. (c) Tenascin expression is strong
around alveolar (al) capillary bed and peripheral
pulmonary veins. AD, alveolar duct; pa, pulmonary artery.
Hall, Hislop, and Haworth: Fetal Development of Human Pulmonary Veins
337
Figure 3. Photomicrographs illustrating (a–e)
development and maturation of pulmonary
venous smooth muscle and (f–k) expression of
ephrinB2 and EphB4 in the developing pulmonary vasculature. (a) Lung at 84 d gestation labeled for -SM actin showing a proximal intrapulmonary vein with a single strongly stained,
thin subendothelial cell layer surrounded by
2–3 additional loosely packed stained flattened
cell layers. Bar: 50 m. (b) Lung at 56 d gestation showing -SM actin expression in some
endothelial cells at the coalescence of two capillaries within the mesenchyme. Bar: 25 m. (c)
Lung at 84 d gestation labeled for -SM actin
in a section adjacent to (a) showing staining of
only the innermost subendothelial layer of
smooth muscle cells. Bar: 50 m. (d) Lung at
140 d gestation labeled for SM1. Expression is
strong in the subendothelial layer of smooth
muscle around small peripheral veins close to
the capillary plexus around the terminal airway buds (TB). Staining of the larger veins
(lpv) beside a lymphatic channel (*) is weak or
absent. Bar: 50 m. (e) Lung at 98 d gestation
labeled for desmin. Section through the extrapulmonary vein (epv) and left atrial wall,
showing desmin staining of the atrial myocardium (amy) and of the end of a layer of myocardial muscle (mm) outside the venous vascular smooth muscle (vsm, desmin negative).
Bar: 50 m. (f ) Aorta at 98 d gestation labeled
for ephrinB2 (no counterstain). It is detectable
in the endothelium of the aorta but not that of
the branch (arrow). Bar: 100 m. (g) Aorta at
98 d gestation labeled for EphB4. No EphB4 is
detectable in the endothelium of the aorta, but
it is strongly expressed in the endothelium of a
small branch (arrow). Bar: 100 m. (h) Lung at
84 d gestation labeled for ephrinB2. It is
strongly expressed in the pulmonary arterial
endothelium and is also detectable in the inner
medial smooth muscle cell layers. Bar: 25 m.
(i) Lung at 84 d gestation labeled for ephrinB2
(frozen section). There is no detectable labeling of the endothelium in an intrapulmonary
vein. Bar: 25 m. ( j) Lung at 84 d gestation labeled for EphB4. Around peripheral airway
generations, expression of EphB4 is strong in
the endothelial cells of the pulmonary vein
(pv), primary capillary plexus (arrowheads),
and pulmonary artery (pa). Bar 50 m. Inset: EphB4 is strongly expressed in the hilar pulmonary vein. Bar: 25 m. (k) Lung at birth
labeled for EphB4 (no counterstain). It is strongly expressed throughout the alveolar capillary bed and in a small vein ( arrow) but is absent from the endothelium of the larger vein into which it drains. Bar: 25 m.
adjacent tissue sections revealed that these desmin-containing cells also contained striated actin filament bundles.
Seeking Differential Identification of Arteries and Veins by
Expression of EphB4 and ephrinB2
In the systemic circulation, the ligand ephrinB2 was strongly
expressed in the aortic endothelium and large coronary arteries of all the embryos and fetuses examined (38–98 d
gestation) but was weaker in small arteries branching from
the aorta (Figure 3f) and in small coronary arteries (the
endocardium was positive). Conversely, the receptor EphB4
was not expressed in the aortic endothelium in any of the
embryos and fetuses examined, but it was expressed in
small arteries branching from the dorsal aorta (Figure 3g).
It was also seen in small coronary arteries and veins (the
endocardium was negative) and other small systemic arteries and veins. Endothelial cells of large cardinal veins
were positive for ephrinB2 but not EphB4.
For the pulmonary circulation, at 38–47 d gestation the
ligand ephrinB2 was expressed in the endothelium of both
338
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 26 2002
TABLE 2
Expression of smooth muscle–specific proteins during muscularization of pulmonary veins during development
-SM Actin
Gestational
Age (d)
38
44–47
56–84
98–140
Neonate
-SM Actin
MHC SM1
Calponin
Hilar–
Hilar–
Hilar–
Hilar–
Midlung Peripheral First Post- Midlung Peripheral
First Post- Midlung Peripheral First Post- Midlung Peripheral First Post(subend
2–3
capillary (subend
2–3
capillary
(subend
2–3
capillary
(subend
2–3
capillary
outer) Generations Generation outer) Generations Generation outer) Generations Generation outer) Generations Generation
†
‡ (
)§ (
) (
)
(
) (
)
(
)
(
)
(
)
Definition of abbreviations: outer, outer medial cell layers; subend, subendothelial cell layer.
*Scoring where medial cells present.
†
, no cells positively stained.
‡
, all cells positively stained.
§
(
), some cells positively stained.
2–3 generation of arteries and veins (up to 25–30 m) and
was not expressed as the vessels increased in size. It colocalized with ephrinB2 in the endothelium of arteries approximately three generations from the capillary bed. Thus,
expression of EphB4 and ephrinB2 did not discriminate, in
humans, between the presumptive venous and arterial endothelia, either pulmonary or systemic.
prospective extrapulmonary and hilar intrapulmonary arteries but not in the peripheral generations or in the capillary plexus (Table 3). By 84–140 d gestation, it was found
in both the endothelium and medial smooth muscle cells
of large proximal intrapulmonary arteries (Figure 3h) and
extended in both to 3–4 generations proximal to the periphery. EphrinB2 was not detectable in the pulmonary
veins at any age before or after birth (Figure 3i). However,
it was found in endothelial cells lining the adjacent lymphatic channels from 98–140 d.
The receptor EphB4 was not detected in the lung buds
before 44 d gestation. At 44–47 d it was expressed in the
endothelium of the prospective extrapulmonary vein (Table 3). In the 56- to 98-d-old fetuses, the endothelium of the
extrapulmonary vein no longer showed EphB4 expression.
However, expression was now strong in all intrapulmonary
veins and arteries and was also detectable in proximal capillaries (Figure 3j and inset). By 105–140 d gestation, strong
expression of EphB4 was found in the endothelial cells of
peripheral veins and arteries and throughout the capillary
plexus around the terminal buds. In the neonatal lungs,
EphB4 was also detected only in respiratory unit veins and
arteries and in the alveolar capillaries (Figure 3k). Thus,
throughout gestation, EphB4 was present in the peripheral
Discussion
The findings in the present study suggest that the pulmonary veins of the human embryo form initially by vasculogenesis from the splanchnopleural mesoderm and that the
later growth in the intraacinar region is likely to be by angiogenesis. At 34 d gestation, serial reconstruction demonstrated physical continuity between the aortic sac, pulmonary artery, peribronchial capillary plexus, pulmonary veins,
and left atrium, suggesting circulation of blood through
the lung bud. This continuity was described previously in
mice at an equivalent gestational age (13). We used EphB4
and ephrinB2 in an attempt to distinguish presumptive
pulmonary veins from pulmonary arteries during vasculogenesis but could not detect either receptor or ligand on
the capillary bed in early embryos. EphrinB2 was expressed on pulmonary arteries but not veins, as described
TABLE 3
Distribution of ephrinB2 and EphB4 in pulmonary arteries and veins: changes with age*
ephrinB2
EphB4
Arteries
Gestational
Age (d)
34–38
44–47
56–84
98–140
Neonate
Veins
Arteries
Veins
ExtraPeripheral
Peripheral
ExtraExtraPeripheral
Peripheral
Extrapulmonary, Mid2–3
Capil2–3
Mid- pulmonary, pulmonary, Mid2–3
Capil2–3
Mid- pulmonary,
Hilar
lung Generations laries Generations lung
Hilar
Hilar
lung Generations laries Generations lung
Hilar
†
‡
* Scoring where lung bud large enough for these vessels to be present.
†
, all cells positively stained.
‡
, no cells positively stained.
§
(
), some cells positively stained.
(
)
(
)§
(
)
Hall, Hislop, and Haworth: Fetal Development of Human Pulmonary Veins
in the systemic circulation of mice (5, 14, 15). However,
EphB4 did not discriminate human pulmonary veins from
arteries (5, 14) and showed temporal and spatial changes
in expression in both structures throughout gestation. Pulmonary venous smooth muscle appeared to be derived
from the mesenchyme by differentiation from undifferentiated mesenchymal precursors. There was no evidence of
smooth muscle derivation from bronchial smooth muscle
as we had previously described in the pulmonary arteries,
probably because the pulmonary veins did not mature in
close proximity to the airways. By mid-gestation, the intrapulmonary veins were closely associated with lymphatic
channels, and together they formed rudimentary septa between the airways. The lymphatic profiles could be important in isolating veins from the influence of airway-derived
vasculogenic or angiogenic signals (8, 16–18).
Origin of Pulmonary Veins
In the human embryo, ephrinB2 and its receptor EphB4
were not early molecular markers of the endothelial cells
within the lung bud. They were expressed sometime after
the cells expressed CD31. Also, as noted above, ephrinB2
and EphB4 expression did not distinguish the endothelium
of presumptive pulmonary arteries and veins. EphB4 labeled the capillaries connected to both arteries and veins.
The discriminatory expression described in systemic vessels in mice is thought to reflect a complex repulsive interaction between receptor and ligand, which is crucial for
the morphogenesis of capillary beds into either arteries or
veins (5). The findings in the present study indicate either
that such repulsion is not the mechanism for arterial and
venous discrimination or that different molecules are responsible for such a mechanism in the human lung.
Our observations suggest that until the end of the
pseudoglandular period, the pulmonary veins originate, like
the arteries, by vasculogenesis from the splanchnopleural
mesenchyme. Induction of vasculogenesis is thought to be
mediated by vascular endothelial growth factor (VEGF),
which is expressed in the epithelium of terminal buds in
the human fetal lung during the late pseudoglandular phase
but not in the adjacent capillary bed (19). In cultured murine embryonic lung, beads impregnated with VEGF stimulate a local vasculogenic response at the leading edge of
branching airways (20). Formation of blood vessels is abnormal and lethal in VEGF-deficient mice (21). In humans, we found that peripheral expansion of the capillary
bed is accompanied proximally by continuous coalescence
and maturation of veins until mid-gestation. Previous studies have shown that all preacinar intrapulmonary veins
have formed by this time (22). After the pseudoglandular
phase, the amount of undifferentiated mesenchyme, the
source of the earlier endothelial tubes, had decreased in
volume. This suggests that endothelial tubes, which form
more peripheral veins, may originate from a different
source, possibly by angiogenesis from existing veins. Ki67labeled replicating cells were identified in venules close to
the terminal bud, and tenascin was newly expressed in the
subepithelium of the terminal buds. Tenascin expression is
known to be abnormally high in angiogenic lesions (11),
and studies on the control of angiogenic sprouting of endothelial cells in vitro have shown that tenascin is essential
339
for the formation of endothelial tubes (8, 9). In mouse
lungs, a shift in the mechanism of vascular expansion corresponds with a decrease in sonic hedgehog expression at
the start of the canalicular phase (23). Transgenic overexpression of sonic hedgehog results in the retention of a
high mesenchymal rate of cell replication and inhibition of
alveolar development (23). In our later fetal and neonatal
cases, EphB4 expression was particularly strong in the
capillary endothelium. This may be associated with a
change from vasculogenic to angiogenic expansion of the
vascular bed. In mouse ephrinB1 and B2, ligands induced
capillary sprouting in vitro with a similar efficiency to
VEGF (14).
Origin and Maturation of Venous Smooth Muscle
Pulmonary venous smooth muscle was derived initially
from undifferentiated mesenchymal cells. No muscle cells
originated from bronchial smooth muscle, as in the pulmonary arteries (2). There were other differences in vascular
smooth muscle development between pulmonary veins
and arteries. The veins were at a greater distance from the
capillary bed and closer to the hilum when they acquired a
smooth muscle coat than were the arteries. Veins acquired
-SM actin positive cells at 56 d gestation, whereas arteries
acquired them from 38 d gestation. Expression of mature
smooth muscle specific proteins in the veins also lagged
behind that seen in the arteries (2). The first structural sign
of smooth muscle cell differentiation in the veins was the
acquisition of an elongated shape by the mesenchymal
cells lying alongside the endothelium. This was associated
with expression of -SM actin. In undifferentiated mesenchymal cells in vitro, elongation is sufficient for the acquisition of differentiated smooth muscle cell characteristics
(24, 25). In the systemic circulation, smooth muscle differentiation from mesenchymal precursors is regulated by
platelet-derived growth factor- released from immature
endothelial cells (26).
In late gestation, the endothelial cells of peripheral veins
expressed -SM actin, as did the pulmonary arteries (2).
Endothelial cells are thought to give rise to smooth muscle
cells in the chick aorta and rat lung (27, 28). Using immunohistochemistry, it is impossible to determine whether
the venous endothelial cells actually give rise to smooth
muscle cells or if this is only a transient change in the phenotype of the endothelial cells that is perhaps regulated by
transforming growth factor expression (17, 29, 30).
Maturation of pulmonary venous smooth muscle was not
a continuous process throughout gestation, as it was in pulmonary arterial smooth muscle (2). The expression of SM-1
disappeared transiently from pulmonary veins greater
than 70–100 m in diameter at 98–140 d gestation. This reduction was associated with an increase in lumen size,
which was maintained even when SM-1 reappeared. This
suggests that developmentally regulated remodeling may
reduce resistance to blood flow. By contrast, developing
pulmonary arteries remained thick walled around a narrow lumen, and SM-1 expression was constant (2). Thus,
regulating expression of specific smooth muscle contractile proteins influenced vascular wall remodeling and presumably pulmonary blood flow in fetal life.
340
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 26 2002
By the end of gestation, all pulmonary venous smooth
muscle had acquired several cytoskeletal proteins characteristic of differentiated vascular smooth muscle. The absence of caldesmon expression is noteworthy because it is
present in the pulmonary artery smooth muscle from 56 d
of gestation (2). Caldesmon is thought to act as a regulator
of contraction in vivo, and tethering of actin to myosin by
caldesmon in vitro blocks actomyosin interaction and so
may inhibit contraction (31). These observations may help
explain why porcine pulmonary veins are more responsive
to contractile agonists than arteries at birth (1).
In conclusion, it appears that the pulmonary veins arise
by vasculogenesis like the pulmonary arteries, but the molecular distinction between presumptive venous and arterial endothelium is still to be determined. The smooth
muscle cells are derived entirely from the surrounding
mesenchyme. Unlike the pulmonary arteries, there is no
initial contribution from the bronchial muscle. In vitro studies
indicate that bronchial epithelium can influence vascular
development (20). The influence may be less in venous
than in arterial development because of a greater diffusion
pathway. The veins are further isolated from airway and
mesenchymal mediators by the lymphatic channels. The
mechanisms regulating pulmonary vascular development
demand an in-depth study including the use of genetically
modified mice. Such experimental studies will improve our
understanding of human development, but experimental
studies must themselves be guided by our present knowledge of human pulmonary vascular development.
Acknowledgment: This study was supported by The British Heart Foundation.
References
1. Arrigoni, F. I., A. A. Hislop, S. G. Haworth, and J. A. Mitchell. 1999. Newborn intrapulmonary veins are more reactive than intrapulmonary arteries
in normal and hypertensive piglets. Am. J. Physiol. 277:L887–L892.
2. Hall, S. M., A. A. Hislop, C. Pierce, and S. G. Haworth. 2000. Prenatal origins of human intrapulmonary arteries: formation and smooth muscle maturation. Am. J. Respir. Cell Mol. Biol. 23:194–203.
3. Bennett, B. D., F. C. Zeigler, Q. Gu, B. Fendly, A. D. Goddard, N. Gillett,
and W. Matthews. 1995. Molecular cloning of a ligand for the Eph-related
receptor protein-tyrosine kinase. Proc. Natl. Acad. Sci. USA 92:1866–1870.
4. Bergemann, A. D., H. J. Cheng, R. Brambilla, R. Klein, and J. G. Flanagan.
1995. ELF-2, a new member for the Eph ligand family, is segmentally expressed in mouse embryos in the region of the hindbrain and newly forming somites. Mol. Cell. Biol. 15:4921–4929.
5. Wang, H. U., Z.-F. Chen, and D. J. Anderson. 1998. Molecular distinction
and angiogenic interaction between embryonic arteries and veins revealed
by ephrin-B2 and its receptor Eph-B4. Cell 93:741–753.
6. Klein, R. 2001. Excitatory Eph receptors and adhesive ephrin ligands. Curr.
Opin. Cell Biol. 13:196–203.
7. Larsen, W. J. 1993. Human Embryology. Churchill Livingstone, New York.
8. Canfield, A. E., and A. M. Schor. 1995. Evidence that tenascin and thrombospondin-1 modulate sprouting of endothelial cells. J. Cell Sci. 108:797–809.
9. Schenk, S., R. Chiquet-Ehrismann, and E. J. Battegay. 1999. The fibrinogen
globe of tenascin-C promotes basic fibroblast growth factor-induced endothelial cell elongation. Mol. Biol. Cell 10:2933–2943.
10. Boyden, E. A. 1977. Development and growth of the airways. In Development of the Lung, Vol. 6. W. A. Hodson, editor. Marcel Dekker, New
York. 3–35.
11. Kostianovsky, M., M. A. Greco, J. Cangiarella, and D. Zagzag. 1997. Tenascin-C expression in ultrastructurally defined angiogenic and vasculogenic
lesions. Ultrastruct. Pathol. 21:537–544.
12. Hungerford, J. E., G. K. Owens, W. S. Argraves, and C. D. Little. 1996. Development of the aortic vessel wall as defined by vascular smooth muscle
and extracellular matrix markers. Dev. Biol. 178:375–392.
13. Schachtner, S. K., Y. Wang, and H. Scott Baldwin. 2000. Qualitative and
quantitative analysis of embryonic pulmonary vessel formation. Am. J.
Respir. Cell Mol. Biol. 22:157–165.
14. Adams, R., G. A. Wilkinson, C. Weiss, F. Diella, N. W. Gale, U. Deutsch,
W. Risau, and R. Klein. 1999. Roles of ephrinB ligands and ephB receptors in cardiovascular development: demarcation of arterial/venous domains, vascular morphogenesis and sprouting angiogenesis. Genes Dev.
13:295–306.
15. Adams, R. H., and R. Klein. 2000. Eph receptors and ephrin ligands: essential mediators of vascular development. Trends Cardiovasc. Med. 10:183–188.
16. Hirschi, K. K., S. A. Rohovsky, and P. A. D’Amore. 1998. PDGF, TGF,
and heterotypic cell-cell interactions mediate endothelial cell-induced recruitment of 10T1/2 cells and their differentiation to a smooth muscle fate.
J. Cell Biol. 141:805–814.
17. Roelen, B. A. J., M. A. Van Rooijen, and C. L. Mummery. 1997. Expression
of ALK-1, a type 1 serine/threonine kinase receptor, coincides with sites of
vasculogenesis and angiogenesis in early mouse development. Dev. Dyn.
209:418–430.
18. Yancopoulos, G. D., S. Davis, N. W. Gale, J. S. Rudge, S. J. Wiegand, and J.
Holash. 2000. Vascular-specific growth factors and blood vessel formation.
Nature 407:242–248.
19. Shifren, J. L., N. Doldi, N. Ferrara, S. Mesiano, and R. B. Jaffe. 1994. In the
human fetus, vascular endothelial growth factor is expressed in epithelial
cells and myocytes, but not vascular endothelium: implications for mode of
action. J. Clin. Endocrin. Metab. 79:316–322.
20. Healy, A. M., L. Morgenthau, X. Zhu, H. W. Farber, and W. V. Cardoso.
2000. VEGF is deposited in the subepithelial matrix at the leading edge of
branching airways and stimulates neovascularization in the murine embryonic lung. Dev. Dyn. 219:341–352.
21. Carmeliet, P., V. Ferreira, G. Breier, S. Pollefeyt, L. Kieckens, M. Gertsenstein, M. Fahrig, A. Vandenhoeck, K. Harpal, C. Eberhardt, C. Declercq,
J. Pawling, L. Moons, D. Collen, W. Risau, and A. Nagy. 1996. Abnormal
blood vessel development and lethality in embryos lacking a single VEGF
allele. Nature 380:435–439.
22. Hislop, A., and L. Reid. 1973. Fetal and childhood development of the intrapulmonary veins in man—branching pattern and structure. Thorax
28:313–319.
23. Bellusci, S., Y. Furuta, M. G. Rush, R. Henderson, G. Winnier, and B. L. M.
Hogan. 1997. Involvement of Sonic hedgehog (Shh) in mouse embryonic
lung growth and morphogenesis. Development 124:53–63.
24. Yang, Y., S. Beqaj, P. Kemp, I. Ariel, and L. Schuger. 2000. Stretch-induced
alternative splicing of serum response factor promotes bronchial myogenesis and is defective in lung hypoplasia. J. Clin. Invest. 106:1321–1330.
25. Relan, N., Y. Yang, S. Beqaj, J. H. Miner, and L. Schuger. 1999. Cell elongation induces laminin 2 chain expression in mouse embryonic mesenchymal cells: role in visceral myogenesis. J. Cell Biol. 147:1341–1350.
26. Hellstrom, M., M. Kalen, P. Lindhal, A. Abramsson, and C. Betsholtz. 1999.
Role of PDGF-B and PDGFR in recruitment of vascular smooth muscle
cells and pericytes during embryonic blood vessel formation in the mouse.
Development 126:3047–3055.
27. DeRuiter, M. C., R. E. Poelmann, J. C. VanMunsteren, V. Mironov, R. R.
Markwald, and A. C. Gittenberger-de-Groot. 1997. Embryonic endothelial cells transdifferentiate into mesenchymal cells expressing smooth muscle actins in vivo and in vitro. Circ. Res. 80:444–451.
28. Mitchell, J. J., J. Woodcock-Mitchell, R. B. Low, K. O. Leslie, and S. E.
Reynolds. 1990. Smooth muscle cell markers in developing rat lung. Am. J.
Respir. Cell Mol. Biol. 3:515–523.
29. Amberger, A., H. Bauer, U. Tontsch, G. Gabbiani, O. Kocher, and H. C.
Bauer. 1991. Reversible expression of sm -actin protein and sm -actin
mRNA in cloned cerebral endothelial cells. FEBS Lett. 287:223–225.
30. Arciniegas, E., A. B. Sutton, T. D. Allen, and A. M. Schor. 1992. Transforming growth factor beta 1 promotes the differentiation of endothelial
cells into smooth muscle-like cells in vitro. J. Cell Sci. 103:521–529.
31. Lee, Y. H., C. Gallant, H. Q. Guo, Y. Li, C.-L. Albert Wang, and K. G.
Morgan. 2000. Regulation of vascular smooth muscle tone by N-terminal
region of caldesmon. J. Biol. Chem. 275:3213–3220.