relationship between vasculogenesis, angiogenesis and … · cv, cardinal vein; g, gizzard; 1,...

Development 105, 473-485 (1989)Printed in Great Britain © The Company of Biologists Limited 1989

473

Relationship between vasculogenesis, angiogenesis and haemopoiesis

during avian ontogeny

LUC PARDANAUD, FOUZIA YASSINE and FRANCOISE DIETERLEN-LIEVRE

Institut d'Embryologie Cellulaire et MoUculaire du CNRS et du College de France, 49 bis Avenue de la Belle Gabrielle, 94736 Nogent surMarne Cedex, France

Summary

Quail-chick intracoelomic grafts of organ rudimentswere used to study the origin of endothelia and haemo-poietic cells during avian organogenesis in conjunctionwith the monoclonal antibody QH1 which recognizes thequail haemangioblastic lineage. Results differed accord-ing to the germ-layer constitution of the grafted rudi-ments. In the case of the limb buds, endothelial cellsfrom the host invaded the graft through an angiogenicprocess. Haemopoietic progenitors from the host alsocolonized the grafted bone marrow.

In contrast, rudiments of internal organs providedtheir own contingent of endothelial precursors, a processtermed vasculogenesis. Nevertheless, haemopoietic cellsin these organs were all derived from the host. In thelung, this extrinsic cell population appeared regularlyscattered around the parabronchi and had a macro-phage-like phenotype. In the pancreas, the granulocyteswhich differentiate as dense aggregates located in thewall of the largest vessels were extrinsic. Similarly in thespleen, a mesodermal primordium that develops in close

association with the pancreatic endoderm, endothelialcells were intrinsic and haemopoietic cells host-derived.

This study demonstrates that, in ontogeny, vasculari-zation obeys different rules depending on which germlayer the mesoderm is associated with: in mesoder-mal/ectodermal rudiments angiogenesis is the rule; inmesodermal/endodermal rudiments, vasculogenesis oc-curs. However, in these internal organs undergoingvasculogenesis, endothelial and haemopoietic cells haveseparate origins.

We put forward the hypothesis that the endoderminduces the emergence of endothelial cells in the associ-ated mesoderm. Formation of blood stem cells may alsoinvolve interactions between endoderm and mesodenn,but in this case the responding capacity of the mesodermappears restricted to the paraaortic region.

Key words: chick, quail, vasculogenesis, angiogenesis,haemopoiesis.

Introduction

By definition, angiogenesis (Hertig, 1935) is the processthrough which new blood vessels arise from preexistingones. Extensively analysed in the case of tumourvascularization (see reviews in Folkman, 1985a and b),this process involves attraction and proliferation ofendothelial cells into tissues which emit appropriatestimuli. The term of vasculogenesis on the other hand isreserved for the emergence of blood vessels occurringde novo in the early embryo (Houser et al. 1961;Gonzalez-Crussi, 1971; Feinberg & Beebe, 1983; Parda-naud etal. 1987; Risau & Lemmon, 1988). Using QH1,a monoclonal antibody specific for quail endothelial andhaemopoietic cells, we have previously described thedevelopment of the endothelial network by vasculogen-esis in the early quail blastodisc (Pardanaud et al. 1987).A similar analysis was carried out by Coffin & Poole(1988).

Scattered experimental evidence demonstrates thatvasculogenesis and angiogenesis can occur as indepen-dent processes during ontogeny. Early authors isolated,removed or destroyed regions from avian blastodiscsand concluded that vascularization developed normallyin the remaining regions, and thus did not rely oningrowth of vessels from removed parts (Miller &McWhorter, 1914; Reagan, 1915). They could show inparticular that vascularization of the area pellucida didnot depend on ingrowth of vessels from the areavasculosa. In quail embryos grafted on a chick extra-embryonic area, we also observed that vessel endo-thelia arose independently in the chick or quail regionsof these 'yolk-sac chimaeras' (Beaupain et al. 1979).Our recent observations clearly indicate that QH1-positive cells arise individually according to a time-specific pattern in different regions of the blastodisc andlink secondarily into a network (Pardanaud et al. 1987).

On the other hand, angiogenesis also occurs later in

474 L. Pardanaud and others

development. These data were established using quail-chick interspecies grafting and subsequent tracing ofquail cells by means of the Feulgen-stained nucleolarmarker (Le Douarin, 1969). In this way, Jotereau & LeDouarin (1978) reached the conclusion that the wholelimb bud, in particular the bone marrow, is colonized byextrinsic endothelial buds. Similarly, Stewart & Wiley(1981) demonstrated that endothelia colonize the brain,while Ekblom etal. (1982) and Sariola etal. (1983,1984)showed that the kidney is invaded by endothelial buds.These data led to the idea that, while the early vasculartree forms by vasculogenesis, organogenesis involves anangiogenic process (Risau & Lemmon, 1988).

We decided to investigate whether an angiogenicprocess was responsible for vascularization of all organrudiments, in particular by comparing the sequence ofevents in organs with an endodermal versus an ectoder-mal epithelial component. We were also interested inthe relationship of haemopoiesis to blood vessel devel-opment. A common precursor for endothelial andhaemopoietic cells in the embryo has been hypothe-sized (Sabin, 1920) and named the haemangioblast(Murray, 1932). In recent years, markers shared bythese two cells types have been described, such asmonoclonal antibodies MB1 (P6ault et al. 1983) andQH1 (Pardanaud et al. 1987). This sharing of anantigenic epitope has been used as an argument for theexistence of a common precursor (Pe"ault etal. 1983). Adescriptive study of the formation of blood islands inthe yolk sac using MB1 also appeared to suggestderivation from the same ancestor (P6ault et al. 1988).

One of us has shown earlier that stem cells respon-sible for definitive haemopoiesis arise in the embryoproper, rather than in the yolk sac (Dieterlen-Lievre,1975, 1984a). Our experiments carried out in vivo or invitro on the origin of intraembryonic stem cells have ledto the conclusion that the paraaortic region of themesoderm in 3- to 4-day (E3-4) avian embryos pro-duces blood stem cells. When the aorta and surround-ing cells were grafted from the quail into the chickdorsal mesentery, the transplants yielded typical foci ofdiffuse haemopoiesis (Dieterlen-Lievre, 1984a and b).The potentialities of cells from chick dissociated aortaewere also tested in clonal cultures (Cormier et al. 1986;Cormier & Dieterlen-Lievre, 1988). This region of theembryo appeared as a rich source of haemopoieticprogenitors. The concentration of progenitors was actu-ally 3 to 4 times higher in aortic wall cell preparationsthan in bone marrow from newborn chickens. Incontrast, cells obtained from E4 embryos deprived oftheir aorta yielded no colonies.

We undertook the present investigation to pinpointthe relative parts played in ontogeny by vasculogenesisand angiogenesis and their relationship to haemopoie-sis. Our aim was to establish the origin of the haeman-gioblastic lineage in rudiments which are derived fromeither the splanchnopleura, i.e. mesoderm plus endo-derm, or from the somatopleura, i.e. mesoderm plusectoderm. Interspecies grafting of these rudiments wascarried out. Cell origins were traced with the QH1antibody, whose affinity is restricted to quail haeman-

gioblastic cells. Thus when the host is a quail, positivecells are extrinsic colonizers while, in chick hosts,positive cells arise from the grafted quail rudiment.

Materials and methods

Quail (Coturnix coturnix japonica) embryos and White Leg-horn chick embryos were used. Donor embryos were incu-bated for 48 to 96h, host embryos 84h, i.e. stage 20 accordingto Hamburger & Hamilton (1951). In order to describe quailembryos stages, the criteria of Hamburger & Hamilton'schick classification were used rather than Zacchei's table(1961) which is less precise with regard to the early stages ofdevelopment.

Wing or leg buds were chosen as mesodermal/ectodermalrudiments. When rudiments were retrieved prior to stage 18HH, the splanchnopleural sheet was separated from thestomatopleural sheet by mechanical dissection in magnesium/calcium-free phosphate buffered saline (PBS). To obtainmesodermal/endodermal rudiments, the donor embryo wasimmobilized left side up by means of two insect needles. Thebody wall was taken off permitting access to lung, gut,pancreas and spleen rudiments which were severed.

E3-5 host embryos were prepared as follows: eggs wereopened through the blunt end and the shell membrane wasripped off after first moistening it with a few drops of PBS.Using two pairs of fine tweezers, the vitelline membrane andthe amnios were carefully removed around the right wing bud.

The explant of one species, marked with a few grains ofsterile animal black, was placed, in the egg, near the hostembryo of the other species. A longitudinal incision was madealong the right wing bud of the host with a sharpened needle.Then the explant was placed on top of the incision and, usingthe needle and a pair of fine tweezers, it was pushed into thehost coelom. The egg was then closed with adhesive Scotchtape and placed at 37 °C.

Operated eggs, supervised daily, were incubated for vari-ous lengths of time after the gTaft. The graft in most cases waslocated on the right side of the body behind the liver, near thelung and the mesonephros. Viscera were cut off and theexplant was carefully removed then cleaned in PBS and fixedin Bouin's fluid for 2 to 3 h (except for the bone rudimentswhich were fixed one or more days depending on their age ofdevelopment). Tissues were embedded in paraffin and seriallysectioned at 7-5 /an.

Sections were rehydrated in PBS following two toluenebaths. The slides were then treated for 20 min with 2-3 % newborn calf serum (Gibco) in PBS in order to saturate non-specific antibody-binding sites. Monoclonal antibody QH1,directly coupled with FTTC, was applied for 45 min at roomtemperature within a humid chamber. The slides were washedtwice in PBS and mounted in PBS. In some cases, a doublelabelling with Hoechst stain (Bisbenzimide H 33258, ServaFeinbiochemica) was performed to visualize nuclei. The slideswere placed in a lmgl"1 solution in PBS for 3 min and thencarefully washed with PBS. They were sealed with nail varnishand observed with a Leitz microscope (Dialux 22) equippedwith u.v. epifluorescence.

Results

Vasculature of rudiments at the time of transplantationQH1 monoclonal antibody has affinity for both endo-thelial and haemopoietic cells in the quail species. The

Endothelial and blood cell relationship in avian embryo 475

Figs 1, 2. Normal development of the vasculature in an E 2-5-3 quail embryo. QH1 immunofluorescence. a, aorta;cv, cardinal vein; g, gizzard; 1, liver; s, spleen; w, vitelline vein; w, wing bud. Bar, 100 /an.Fig. 1. Stage 17HH. Cross section at the wing bud level. The main vessels are present, displaying a QHl-positiveendothelium. Some vascular profiles are present at the base of the limb bud (arrows) while the distal part is devoid ofpositive cells.Fig. 2. Stage 19HH. Cross section through the trunk with dorsal side on the left. QHl-positive cells have not yet entered thespleen rudiment. The vascular network in the limb bud is still poorly developed. The ventral aspect of the aorticendothelium is conspicuously lined with haemopoietic cells (arrow), a frequent observation at E3-E4.


appearance of these two cell types is very different, sothat QHl-positive cells can be readily diagnosed as oneor the other type. Obviously the interpretation of theresults in interspecies grafts rests on the state ofvascularization at the time of transplantation. Normalquail embryos stained with QH1 between stages 17 and21HH served to monitor the state of vasculature indifferent rudiments at stages used to obtain donorrudiments. At stage 17 (Fig. 1), the vascular treecomprised the main vessels, aorta, cardinal veins,vitelline veins and perineural vasculature. The base ofthe wing bud displayed several endothelial profileswhile its distal mesoderm appeared unvascularized. Thesplanchnopleura had not completed the morphogeneticmovements closing the gut and the rudiments of inter-nal organs could not yet be identified as buds. Well-developed endothelial profiles were present betweenthe mesodermal and the endodermal layers of thesplanchnopleura. At stage 19 (Fig. 2), the gut was wellvascularized, while the spleen rudiment contained nopositive endothelia. The first endothelial profiles pen-etrated into this rudiment at stage 21. What should bekept in mind from this brief description is that, apartfrom the spleen, all the explanted rudiments werealready partly vascularized at the time of explantation.

Haemangioblastic patterns in grafted limb buds(Table 1)

Chick rudiments in quail hostsAll vessels were entirely lined with QHl-positive endo-thelium, independent of the age at explantation and theage at autopsy of the transplants. A fluorescent networkinvested the whole exflant, in particular connectivetissue surrounding cartfege (Fig. 3, 5), perichondrium(Fig. 6), muscular masses (Fig. 7) and feather buds(Fig. 8). Cartilage appeared as avascular islands devoidof fluorescent cells. Apart from the fluorescentnetwork, these explants were pervaded with a greatnumber of isolated QHl-positive cells. The QHl-positive cells located in the connective tissue around thecartilage displayed two different appearances. Roundcells, often in groups, bearing a homogeneous surfaceQH1 labelling, evoked haemopoietic precursor cells.Cells of the other type, which were wheel-shaped andlarger, might be macrophages.

In the older explants (Ell to E13), QHl-positiveendothelial cells were the first to invade the marrow(Fig. 9). When the marrow was well developed, thewhole endothelial network was positive (Figs 9-11).Positive vessels were also present in bone lacunae (Figs

Figs 3-8. Haemangioblastic patterns in non-osseous tissueof experimental limb buds. QH1 immunofluorescence.c, cartilage; p, perichondrium.Fig. 3. Chick limb bud in quail host (st 18HH until E12).QHl-positive vessels are present in the connective tissue(arrows) and the perichondrium (thin arrow) around thetwo cartilaginous masses. QHl-positive isolated cells arealso observed in the connective tissue (arrowhead). Bar,100 /an.Fig. 4. Quail limb bud in chick host (st 16HH until E13).The non-cartilaginous tissue is completely devoid of positivecells. Compare with the reverse combination shown inFig. 3. Bar, 100 /an.Fig. 5. Chick limb bud in quail host (st 18HH until E13).The vascular plexus depicted here is typically observed inthe connective tissue between cartilaginous rudiments ofbones; fusiform (arrow) or round (arrowhead) isolated cellsare also present. The different shapes indicate that theformer are endothelial and the latter haemopoietic. Bar,50 /an.Fig. 6. Chick limb bud in quail host (st 18HH until E13).The perichondrium, well developed at this level, is nowinvaded by endothelia (arrow). Numerous positivehaemopoietic cells are present in the connective tissue(arrowheads). Bar, 50/an.Fig. 7. Chick limb bud in quail host (st 18HH until E12).Two muscular masses are present, one cross-sectioned (t),the other longitudinally sectioned (1). Delicate positiveendothelia are located between and around myotubes. Bar,100/an.Fig. 8. Chick limb bud in quail host (st 18HH until E13).Three feather buds are present, two cross-sectioned, onelongitudinally sectioned. In their shaft, positive endotheliasurround the pulp (arrows). A bud emerges with itsvascular ring at the base of the longitudinally sectionedfeather. Note that while the smooth muscles of thelongitudinal feather are poorly vascularized, positiveendothelia and haemopoietic cells are numerous in theconnective tissue (arrowhead). Bar, 100/an.

9, 10). Isolated, round positive cells were observed ingreat numbers in the extravascular compartment ofbone marrow (Figs 10, 11): they could be identified asgranulocytes on the basis of their coarse intracytoplas-mic granules. The vascular compartment containednegative mature erythrocytes and rare QHl-positivecells, devoid of granules (Fig. 11). At the periphery ofthe marrow, close to the bone, a few elongated verylarge positive cells were present (Fig. 10). WithHoechst nuclear stain these cells usually appearedplurinucleated, a characteristic feature of osteoclasts(Fig. 12A,B). The bone was lined with a conspicuousmonolayer of QH1-negative cells. The osteoclasts could

Table 1. QHl-positive cells in grafted limb buds

WingLeg

* Explantedt Explanted

betweenbetween

Endothelia

stages 16-21HHstages 16-21HH

Chick in quail*

Blood cells Explant number

+ 5+ 4

, examined between E9-E13., examined between E10-E17.

Endothelia

-

Quail

Blood

-

in chickt

cells Explant number

1113


cling directly to the bone, or could be located on theother side of the negative monolayer.

Quail rudiments in chick hostsThe great majority of vessels was completely negative inimmunofluorescence (Fig. 4). QHl-positive endothelialcells were present but their scarcity was remarkable inthese transplants, whether they were wing or leg. The

younger the buds at explantation, the less numerouswere these positive cells. In most cases, QHl-positivecells were isolated and located at random within theexplants; they had an elongated appearance and dis-played pseudopods. Exceptionally, QHl-positive cellscould be integrated in a chimaeric vessel in particular inbuds retrieved beyond stage 18HH. No positive haemo-poietic cells or osteoclasts were ever seen in therudiment.


Figs 9-12. Haemangioblastic patterns in experimental chick bone marrow (chick limb bud, st 18 until E13, in quail host).QH1 immunofluorescence.Fig. 9. Cross-sectioned bone rudiments show three different stages of marrow development (1,2,3). As the marrowcompletely replaces the cartilage, the haemopoietic cell population becomes more prominent. In the bone lacunae, positivevessels are visible (arrows). In the connective tissue, an artery displays a chimaeric endothelium (arrowhead). Bar, 100 fan.Fig. 10. In marrow and bone lacunae, thin positive endothelia delineate lumina loaded with negative mature erythrocytesand occasional positive cells. In the marrow extravascular compartment, positive granulocytes are abundant. At theperiphery of the marrow, large fusiform positive osteoclasts sit directly against the bone (arrows) or on the other side of anegative cell monolayer (arrowheads). Bar, 50 fim.Fig. 11. Higher magnification of the marrow. Vessels lined by thin positive endothelia contain negative mature erythrocytes(e). Positive cells (probably erythroblasts) are scarcely present in the lumen of vessels (arrow). The extravascularcompartment is filled with positive granulocytes. Bar, 25/an.Fig. 12. Double labelling of an osteoclast. (A) QH1 staining: The binucleate positive cell displays pseudopods and is in closecontact with bone matrix (b) under the negative cell monolayer lining the bone which is barely distinguishable. (B) Hoechstnuclear staining clearly shows the two nuclei of the osteoclast (arrows) and the nuclei of the QHl-negative cell monolayer(framed in the stippled line). The bone matrix is devoid of nuclei. Bar, 20 fan.

Haemangioblastic patterns in grafted internal organrudiments (Table 2)Whereas mesodermal/ectodermal rudiments under-went an all-or-none colonization by immunofluorescentcells, two QHl-positive populations were resolved inmesodermal/endodermal transplants, one extrinsic andone intrinsic.

Chick rudiments in quail hostsIn this combination, endothelia were not labelled;

rather, numerous scattered or aggregated cells belong-ing to the haemopoietic lineage were QH1 positive,with different phenotypes and distributions dependingon the rudiment. For instance, in the lung rudiments,most vessels had a QHl-negative endothelium; onlyvery scarce positive endothelial cells could be seen, andthey were usually isolated, while positive vascularprofiles were small and exceptional (Fig. 13A,B)-Regularly scattered positive cells were prominentwithin the mesenchyme: these cells were round, had a


Table 2. QHl-positive cells in mesodermaljendodermal grafts

Stomach/intestinePancreasLung

* Explanted betweent Explanted between

stagesstages

Endothelia

15-18HH,15-21HH,

Chick in quail*

Blood cells Explant number

+ 7+ 4+ 4

examined between E8-E13.examined between E9-E17.

Endothelia

+ +

+

Quail

Blood

—

in chickt


66

11

brilliantly positive cytoplasm and were distributed uni-formly everywhere around the parabronchi(Fig. 13A,B). Occasionally they were gathered intolarge aggregates in the vicinity of blood vessels. Thesecells were obviously different in shape and size from thefew positive endothelial cells. Their host origin, widedistribution and general occurrence in all explantsindicate they probably colonize the lung through aregular process.

QHl-negative endothelium lined all vessels in thepancreas as well (Figs 14, 21). Prominent accumu-lations of round positive cells surrounded the largervessels. Such cells, closely aggregated, were in mostcases restricted to the vicinity of these vessels (Figs 14,21). In the case of gut rudiments, six out of sevenexplants displayed a completely QHl-negative endo-thelial network (Fig. 15). The last graft, transplanted atstage 16HH and harvested at E10-5, exhibited scarceQHl-positive cells inserted in some endothelia; a fewsmall endothelial profiles were also entirely QHl posi-tive. On the other hand, in all explants a prominent cellpopulation was recognized by QHl antibody; it wasmade of round, uniformly marked cells, dispersed inthe wall of the gut (Fig. 15). These cells, presumablyhaemopoietic cells, were rather numerous in two of theexplants, in particular at the basal part of the villi. Inthe five others, haemopoietic cells were less numerousor even scarce.

Quail rudiments in chick hostsIn this combination, endothelia were QHl positivewhile no other positive cells were present.

In the lung rudiments, QHl-positive endotheliumbuilt a polyhedral network around the parabronchi(Fig. 16A,B). In two rudiments, among the oldest atexplantation, a few vessels displayed a QHl-negative orchimaeric endothelium. QHl-positive cells with a hae-mopoietic appearance were completely absent from allexplants. It should be noted that parabronchial endo-

derm in the quail always appeared weakly QHl posi-tive.

In pancreatic grafts, the whole vascular networkdisplayed a QHl-positive endothelium (Figs 17, 22). Inone rudiment, explanted at stage 21HH, scarce vesselswith a QHl negative or chimaeric endothelium wereencountered. No QHl-positive haemopoietic cells wereever present. As previously observed for the lung andpancreas, the differentiated gut tissue, whether gizzard-or intestine-like, possessed an entire complement ofvessels with a QHl-positive endothelium. In the gutwall, longitudinal vessels were located as two concentricrings, an internal one encircling the basal part of thevilli and an external one at the outside rim of themucosa. Radial vessels linked the two rings (Fig. 18). Inone exceptional graft explanted at stage 21HH, i.e.comparatively late, a few vascular profiles had either aQHl-negative or chimaeric endothelium. In no rudi-ments were QHl-positive haemopoietic cells observed.When compared to the lung endoderm, it is noteworthythat the gut endoderm is not stained by QHl.

Haemangioblastic patterns in grafted spleen (Table 3)The pattern of QHl cell development was similar in thismesodermal organ to that observed for mesoderma-1/endodermal ones.

Chick rudiments in quail hostsWhatever the schedule of transplantation, homo-geneous results were obtained. Monoclonal antibodyQHl recognized densely aggregated round cells distrib-uted all over the developing organ (Fig. 19A). It wasdifficult to diagnose these positive cells as haemopoieticor endothelial. However when large vessels, eitherveins or arteries, were sectioned it was clear that theirendothelium was QHl negative (Fig. 19B).

Quail rudiments in chick hostsA fine-meshed network appeared QHl positive in thiscombination. The overall appearance of this network

* Explantedt Explanted

Endothelia

betweenbetween

stagesstages

TableChick in quail*

3. QHl-positive cells in spleen

Blood cells Explant number Endothelia

19-36HH,17-26HH,

examinedexamined

31 +

between E6-E15.between E9 and hatching.

graftsQuail

Blood

in chickt


56


1 7 , • ' "

was not radically different from that of positive cellaggregates in the reverse combination, i.e. chick intoquail (Fig. 20A). However, when profiles of veins orarteries were present in the sections, their endothelium

appeared QH1 positive (Fig. 20B). Only a minority oftransplanted spleens presented scarce vessels with aQHl-negative or chimaeric endothelium.

From these observations on larger vessels, as well as


Figs 13-18. Haemangioblastic patterns in experimentalrudiments of internal organs. QH1 immunofluorescence.Fig. 13. A and B. Chick lung rudiment in quail host (st15HH until Ell). (A) Numerous QHl-positive round cellsare distributed around the parabronchi (b). A very smallgroup of positive endothelial profiles is present (arrows).Bright blobs are mature erythrocytes with a non-specificyellow fluorescence. Bar, 100 fan. (B) Higher magnificationshowing positive haemopoietic cells in the pulmonarymesenchyme in the vicinity of two large vessels. The latterare lined with a negative endothelium and contain severalpositive haemopoietic cells in their lumen (arrows). A smallvessel with chimaeric endothelium is present (arrowhead).Bar, 50 fan.Fig. 14. Chick pancreatic rudiment in quail host (st 20HHuntil E13). An abundant positive haemopoietic population(arrowhead) is aggregated around vessels whoseendothelium is negative (arrow). Bar, 100 fan.Fig. 15. Chick gut rudiment in quail host (st 19HH untilE13). No positive vascular profile is observed in the gutwall. Positive haemopoietic cells are present at the base ofvilli (arrow). To the left, part of the pancreatic rudiment(p) displays the characteristic dense positive population ofhaemopoietic cells. Bar, 100/an.Fig. 16. A and B. Quail lung rudiment in chick host (st21HH until E12). The parabronchi (b) are surrounded byvascular profiles with positive endothelia. No positivehaemopoietic cells are present (compare with Figs 13A andB). It may be noted that parabronchial endoderm is weaklystained by QH1 antibody (arrow). (A) Bar, 100 fan.(B)Bar, 50 fan.Fig. 17. Quail pancreatic rudiment in chick host (st 17HHuntil E13). Numerous vascular profiles with positiveendothelia are observed. No positive haemopoietic cells arepresent (compare with Fig. 14). Bar, 100 fan.Fig. 18. Quail gut rudiment in chick host (st 17HH untilE13). Two rings of longitudinal positive vascular profiles arepresent (arrows), one at the base of the villi, the other atthe external border of the gut wall. These rings areconnected together by transverse vessels (arrowhead). Nopositive haemopoietic cells are present, p, pancreas. Bar,100 fan.

from comparison with results concerning other internalorgans (i.e. pancreas, Figs 21 and 22), we feel able toconclude that QHl-positive cells in chick splenic rudi-ments grown in quail hosts were haemopoietic, while inquail rudiments grafted to chick hosts QHl-positivecells correspond to the sinusoidal network.

Discussion

Using quail/chick interspecies transplants and analys-ing grafted tissues by means of a quail haemangioblasticspecific monoclonal antibody, we have been able todistinguish two different processes responsible for en-dothelial development during organogenesis in theavian embryo. Rudiments composed of mesoderm andectoderm are sites for angiogenesis, i.e. invasion byendothelial cells or their precursors, while mesoder-mal/endodermal rudiments undergo vasculogenesis,i.e. differentiation of endothelial cells from selfcon-tained mesodermal precursors. Furthermore, our ex-

periments have uncoupled haemopoiesis from vasculo-genesis while also showing that, in the bone marrow,angiogenesis and haemopoiesis both proceed fromextrinsic progenitors. Thus, in this tissue, endothelialcells and blood cells may or may not descend from acommon precursor.

The origin of endothelia, haemopoietic cells andosteoclasts in the bone marrow has been investigatedearlier using the same approach as ours (Jotereau & LeDouarin, 1978). The novel feature in our experiments isthat we analysed the derivations of the cell lineages ofinterest using monoclonal antibody QH1 instead ofrelying on Feulgen staining of the quail nucleolarmarker. With the very sensitive detection provided bythe monoclonal antibody, we confirm that endothelialand haemopoietic cells and osteoclasts are all extrinsicto the rudiments. QH1 makes the identification of allthese cells extremely precise, whereas Jotereau and LeDouarin found that pericytes and endothelial cells werenot always distinguishable from one another on Feul-gen-stained preparations and that only 60% of quailosteoclasts displayed the nucleolus-associated hetero-chromatin condensations. Nevertheless Jotereau andLe Douarin interpreted their observations as indicatingan extrinsic origin of osteoclasts and haemopoietic cellsand of the whole endothelial network. Theirs was thefirst experimental demonstration that osteoclasts be-longed to the haemopoietic lineage. With QH1 weconfirm that the extrapolation made by these authors toall cells of these types was justified. We would like tostress that, in our experiments, the explantations wereperformed earlier than in Jotereau and Le Douarin'swork, at stages 16 to 21 instead of at stages 23 to 25; thisapproximately one-day difference did not make anydifference in the results.

All investigators who have described the develop-ment of the limb bud vasculature have clearly shownthat a capillary plexus is present as soon as the limb budbegins protruding from the body wall, i.e. stage 17HH(Searls & Janners, 1971), despite the existence of amarginal zone of mesoderm devoid of capillaries(Romanoff, 1960; Wilson, 1983; Drushel etal. 1985). Inour preparations, a few endothelial profiles werealready present at stage 16HH. The contribution ofthese cells to the vascular network of the developedlimb was always marginal, if not negligible. The samewas true in Jotereau and Le Douarin's experiments inwhich transplanted limb buds were one day older. Thusit appears that these early colonizing endothelial budsundergo little proliferation in the limb environment.Regression of the limb bud vasculature as cartilagedifferentiates has been described (Feinberg et al. 1986;Latker et al. 1986; Hallmann et al. 1987). Necrosis ofendothelial cells at this period may be involved in thisregression. This process would explain the restrictedparticipation of early colonizing endothelial cells in thelater vascular network.

Jotereau & Le Douarin (1978) have shown thatendothelial cells still colonize a femur rudiment fromE8-13 quail embryos or from E9-15 chick embryosgrafted into the opposite species, and that both endo-


thelial and haemopoietic cells from the two species areusually mixed in such experimental bone marrows. Ourfuture experiments will aim to establish the relationshipbetween these two cell lineages in the bone marrow andto specify whether the colonization of the limb bud is acontinuous process or whether separate waves of endo-thelial buds grow into the developing limb.

Angiogenesis has also been demonstrated in the caseof the brain, the thymus and the kidney. In the case ofbrain (Stewart & Wiley, 1981), the transplants wereobtained from stage 13HH quails and grafted to thecoelom of E3 chick embryos. According to the authors,'the ectoderm and a small amount of mesoderm wereleft adhering to the brain fragments'. This mesoderm


Figs 19-22. Haemangioblastic patterns in experimentalspleen and pancreas. QH1 immunofluorescence.Fig. 19. A and B. Chick spleen rudiment in quail host(st 19HH until E13). (A) General view. Bar, 100/an.(B) Higher magnification. Bar, 50 /an. Numerous QH1-positive round cells are distributed all over the organ. In Ba vein (v) and an artery (a) are depicted. In both theendothelium is QH1 negative. Positivity in the artery is dueto non-specific fluorescence of red cells.Fig. 20. Quail spleen rudiment in chick host (st 17HH untilE13). (A) General view, p, pancreas. Bar, 100/an.(B) Higher magnification. Bar, 50/an. Positive cells liningthe sinusoids build a delicate network, readilydistinguishable from the contiguous positive cells of thereverse combination (Fig. 19B). Furthermore, arteries (a)are lined with positive endothelium. Comparison ofFig. 20A with Fig. 19A demonstrates that grafted quail andchick rudiments grow to a species-characteristic size in theforeign host.Fig. 21. Chick pancreatic rudiment in quail host (st 19HHuntil E13). Numerous positive haemopoietic cells surroundnegative vessels. These cells are similar in phenotype to theones in spleen from Fig. 19B. Bar, 50/an.Fig. 22. Quail pancreatic rudiment in chick host (st 17HHuntil E13). The thin vascular profiles are similar to, buteasier to identify than, the ones present in the spleenrudiment of Fig. 20B. Bar, 50/an.

Comparison of Figs 21 and 22 emphasizes the differencesin the two cell populations examined with this experimentalmodel. Their clear cut identification in the pancreas makesit easier to recognize the corresponding cell lineages in theexperimental spleen of Figs 19B and 20B.

did not give rise to endothelial cells detectable in thequail-chick marker system. In the experiments involv-ing the thymus (Le Lievre & Le Douarin, 1975), thequail mesencephalic neural crest (and neural tube) wassubstituted for that of the chick host. The mesenchymeof the thymus was derived from the neural crest, i.e. theso-called mesectoderm, in accordance with the origin ofthe mesenchyme in this whole region; the thymicmesenchyme, including pericytes, was quail. In con-trast, endothelial cells were chick, thus derived fromthe host. This is, to our knowledge, the only clearcutcase where pericytes and endothelial cells were shownto have a separate origin. The authors concluded thatthe mesectoderm, in contrast with lateral plate meso-derm, did not have the potentiality to give rise toendothelium. Finally, kidney mesenchyme has beenshown to be unable to give rise to endothelial cells(Ekblom etal. 1982; Sariola etal. 1983,1984). Naturallyor artificially induced mouse metanephros was trans-planted on the quail chorioallantoic membrane, whereit became vascularized by quail endothelial buds. Ek-blom et al. (1982) could also demonstrate angiogenesisoriginating from the chick host when quail kidneyrudiments were placed on the chick chorioallantois.

These various rudiments, in which angiogenesis is therule, involve a neurectodermal rudiment (brain), amesodermal/ectodermal rudiment (limb bud), a mesec-todermal/endodermal rudiment (thymus) or a purelymesodermal rudiment (kidney). On the other hand, wenow report evidence for vasculogenesis in various

mesodermal/endodermal rudiments and in one meso-dermal rudiment, the spleen. It could be argued thatendothelial cells or their precursors have already col-onized these rudiments prior to transplantation. How-ever (1) our descriptive and experimental data favour insitu emergence of endothelial cells in the whole splanch-nopleura (Pardanaud etal. 1987; Beaupain etal. 1979);(2) the occurrence of vasculogenesis or angiogenesis isso clearly contrasted in splanchnopleural versus soma-topleural rudiments that entirely different mechanismsseem to be involved in the vascularization of these germlayers; (3) in one of the rudiments undergoing vasculo-genesis (spleen), QHl-positive cells are not presentprior to transplantation. Our hypothesis is that endo-derm plays a major role in the emergence of thehaemangioblastic lineage in associated mesoderm. Allthe rudiments in which our experiments demonstratevasculogenesis have an endodermal component, includ-ing the spleen which develops in close contact with thepancreas. The exception to this rule would seem to bethe case of the thymus, which undergoes angiogenesisdespite the endodermal nature of the thymic epi-thelium. Le Lievre and Le Douarin's interpretationaccording to which the mesectoderm is devoid ofendothelium-forming potentiality fits the data. It is alsopossible that ectoderm exerts a negative effect on theemergence of endothelial cells in associated mesoderm.In that respect, it is interesting to recall that Kessel &Fabian (1987) have described an inhibitory influence ofectoderm from early blastodiscs on mesodermal eryth-ropoiesis.

The hypothesis according to which endoderm plays aprimordial role in the emergence of the haemangioblas-tic lineage in the mesoderm is not new, having beenadvocated by Wilt (1965) for the formation of bloodislands in the yolk sac. Miura & Wilt (1969, 1970)obtained convincing evidence for the positive roleplayed by endoderm on this process in dissociation-reassociation experiments. In the absence of endoderm,only small collections of blood cells formed and theylacked any endothelial envelope. Endoderm exerted itseffect on mesoderm in transfilter association, thusindicating a mediation by a diffusible substance. How-ever, Miura and Wilt could not be sure whether thiseffect was a true induction or a trophic effect.

In the chimaeric splanchnopleural rudiments, sur-prisingly, endothelial cells were intrinsic and haemopoi-etic cells were not. This is particularly striking in thecase of the spleen, a truly haemopoietic organ in theavian embryo, that harbours 10 times more monocyte/macrophage progenitors (M-CFC) than the newbornbone marrow at the peak of its activity from El l to E13(Yassine & Dieterlen-Lievre, unpublished). Another ofthe splanchnopleural anlagen which becomes colonizedby extrinsic blood cell progenitors is the pancreas, anactive granulopoietic organ during chick development(Benazzi-Lentati, 1932; Dieterlen-Lievre, 1965). Thespleen and pancreas rudiments both develop in thedorsal mesentery, in which haemopoietic stem cellsemerge from E3 to E5 (Dieterlen-Lievre & Martin,1981; Dieterlen-Lievre, 1984a; Cormier et al. 1986;


Cormier & Dieterlen-Lievre, 1988). Yet splenic andpancreatic stroma do not give rise to their own progeni-tors.

Two comments should be made concerning the lung.The first concerns the haemopoietic cells that invade it.The lung is considered as a site of local immuneresponses involving interactions between T and Blymphocytes and macrophages (reviewed in Bienen-stock, 1984). Macrophages are a particularly prominentpopulation in the immune pulmonary functions(Winkler, 1988). The QHl-positive cells that settle inthe chick embryonic lung grafted to the quail probablybelong to this population, because, at the time they areobserved, seeding of lymphocytes in secondary lymph-oid organs has not begun. Future analysis will beapplied to the characterization of these cells. The othercomment concerns the fact that the quail pulmonaryendoderm appears weakly QHl-positive. A relatedobservation has been made with monoclonal antibodyMB1 which has a similar affinity for the quail haeman-gioblastic lineage (P6ault et al. 1983). In chimaericbursa of Fabricius, it was found that when a non-lymphoid quail bursal rudiment was grafted into a chickhost, the bursal endoderm started expressing MB1reactivity precisely when colonizing B progenitors camein contact with it (Belo et al. 1985). It might beworthwhile to investigate whether QH1 reactivity in thelung endoderm is related to the entry of haemopoieticcells.

Since the endothelium-forming capacity of the meso-derm depends on its germ layer association, it istempting to speculate that growth factors synthesized,respectively, by the endoderm and ectoderm may beresponsible for vasculogenesis or angiogenesis. Anendothelial cell growth factor related to human aFGFhas been isolated from embryonic and adult chick brain(Risau, 1986; Risau et al. 1988). Future research willfocus on the potential involvement of this and othergrowth and chemoattractive factors in vasculogenesisand angiogenesis.

We gratefully acknowledge the efficient technical assistanceof Michele Klaine and Dominique Duchatel, thank YannRantier for the photography and Marie-Francoise Meunierfor typing the manuscript. We are grateful to Dr EvanBalaban for critical reading of the manuscript.

References

BEAUPAIN, D., MARTIN, C. & DIETEKLEN-LIEVRE, F. (1979). Aredevelopmental hemoglobin changes related to the origin of stemcells and site of erythropoiesis? Blood 53, 212-225.

BELO, M., MARTIN, C , CORBEL, C. & LE DOUARIN, N. M. (1985).A novel method to bursectomize avian embryos and obtain quail-chick bursal chimaeras. I. Immunocytochemical analysis of suchchimaeras by using species specific monoclonal antibodies. /.Immunol. 135, 3785-3794.

BENNAZZI-LENTATI, G. (1932). Sul focolaio eosinofilopoietico delpancreas di Uccelli. Monitore Zool. Ital. 43, 115-118.

BIENENSTOCK, J. (1984). The lung as an immunologic organ. Ann.Rev. Med. 35, 49-62.

COFFIN, J. D. & POOLE, T. J. (1988). Embryonic vasculardevelopment: Immunohistochemical identification of the origin

and subsequent morphogenesis of the major vessel primordia inquail embryos. Development 102, 735-748.

CORMIER, F. & DIETERLEN-LIEVRE, F. (1988). The wall of the chickembryo aorta harbours M-CFC, G-CFC, GM-CFC and BFU-E.Development 102, 279-285.

CORMIER, F., DE PAZ, P. & DIETERLEN-LIEVRE, F. (1986). In vitrodetection of cells with monocytic potentiality in the wall of thechick embryo aorta. Devi Biol. 118, 167-175.

DIETERLEN-LIEVRE, F. (1965). Etude morphologique etexpe r̂imentale de la differentiation du pancrias chez l'embryonde Poulet. Bull. Biol. Fr. Belg. 94, 1-116.

DIETERLEN-LIEVKE, F. (1975). On the origin of haemopoietic stemcells in the avian embryo: an experimental approach. J.Embryol. exp. Morph. 33, 607-619.

DIETERLEN-LIEVRE, F. (1984a). Emergence of intraembryonic bloodstem cells in avian chimeras by means of monoclonal antibodies.Devi. comp. Immunol. 3, 75-80.

DIETERLEN-LIEVRE, F. (19846). Blood in chimeras. In Chimeras inDevelopmental Biology, Cnp. 6, pp. 133-163. London: AcademicPress.

DIETERLEN-LIEVRE, F. & MARTIN, C. (1981). Diffuse intraembryonichemopoiesis in normal and chimeric avian development. DeviBiol. 88, 180-191.

DRUSHEL, R. F., PECHAK, D. G. & CAPLAN, A. I. (1985). Theanatomy, ultrastructure and fluid dynamics of the developingvascularure of the chick wing bud. Cell Diff. 16, 13-28.

EKBLOM, P., SARIOLA, H., KARKINEN, M. & SAXEN, L. (1982). Theorigin of the glomerular endothehum. Cell Diff. 11, 35-39.

FEINBERG, R. N. & BEEBE, D. C. (1983). Hyaluronate invasculogenesis. Science 220, 1177-1179.

FEINBERG, R. N., LATKER, C. H. & BEEBE, D. C. (1986). Localizedvascular regression during limb morphogenesis in the chickenembryo: I. Spatial and temporal changes in the vascular pattern.Anat. Rec. 214, 405-409.

FOLKMAN, J. (1985a). Tumor angiogenesis. Adv. Cancer Res. 43,175-203.

FOLKMAN, J. (19856). Toward an understanding of angiogenesis:Search and discovery. Perspectives in Biology and Medicine 29,10-36.

GONZALEZ-CRUSSI, F. (1971). Vasculogenesis in the chick embryo.An ultrastructural study. Am. J. Anat. 130, 441-460.

HALLMANN, R., FEINBERG, R. N., LATKER, C. H., SASSE, J. &RISAU, W. (1987). Regression of blood vessels precedes cartilagedifferentiation during chick limb development. Differentiation 34,98-105.

HAMBURGER, V. & HAMILTON, H. L. (1951). A series of normalstages in the development of the chick embryo. /. Morph. 88,49-92.

HERTIG, A. T. (1935). Angiogenesis in the early human chorion andin the primary placenta of the macaque monkey. Contnb.Embryol. Carnegie Inst. Washington 25, 37-81.

HOUSER, J. N., ACKERMAN, G. A. & KNOUFF, R. A. (1961).Vasculogenesis and erythropoiesis in the living yolk sac of thechick embryo. A phase microscopy study. Anat. Rec. 140, 29-43.

JOTEREAU, F. V. & LE DOUARIN, N. M. (1978). The developmentalrelationship between osteocytes and osteoclasts: a study using thequail-chick nuclear marker in endochondral ossification. DeviBiol. 63, 253-265.

KESSEL, J. & FABIAN, B. (1987). Inhibitory and stimulatoryinfluences on mesodermal erythropoiesis in the early chickblastoderm. Development 101, 45-49.

LATKER, C. H., FEINBERG, R. N. & BEEBE, D. C. (1986). Localizedvascular regression during limb morphogenesis in the chickenembryo: II. Morphological changes in the vasculature. Anat.Rec. 214, 410-417.

LE DOUARIN, N. M. (1969). Particularitis du noyau interphasiquechez la Caille japonaise (Coturnix coturnix japonica). Utilisationde ces particularity comme "marquage biologique" dans lesrecherches sur les interactions tissulaires et les migrationscellulaires au cours de Pontog6nese. Bull. Biol. Fr. Belg. 103,435-452.

LE LIEVRE, C. & LE DOUARIN, N. (1975). Mesenchymal derivativesof the neural crest: analysis of chimaeric quail and chick embryo.J. Embryol. exp. Morph. 34, 124-154.


MILLER, A. M. & MCWHORTER, J. E. (1914). Experiments on thedevelopment of blood vessels in the area pellucida andembryonic body of the chick. Anat. Rec. 8, 203-227.

MIURA, Y. & WILT, F. H. (1969). Tissue interaction and theformation of the first erythroblasts of chick embryo. Devi Biol.19, 201-211.

MIURA, Y. & WILT, F. H. (1970). The formation of blood-islands indissociated-reaggregated chick embryo yolk sac cells. Expl CellRes. 59, 217-226.

MURRAY, P. D. F. (1932). The development in vitro of the blood ofthe early chick embryo. Proc. Roy. Soc. (London) 11, 497-521.

PARDANAUD, L., ALTMANN, C , Krros, P., DIETERLEN-LIEVRE, F. &BUCK, C. A. (1987). Vasculogenesis in the early quail blastodiscas studied with a monoclonal antibody recognizing endothelialcells. Development 100, 339-349.

PEAULT, B., COLTEY, M. & LE DOUARIN, N. M. (1988). Ontogenicemergence of a quail leukocyte/endothelium cell surface antigen.Cell Diff. 23, 165-174.

PEAULT, B., THIERY, J. P. & LE DOUARIN, N. M. (1983). A surfacemarker for the hemopoietic and endothelial cell lineage in thequail species defined by a monoclonal antibody. Proc. natn.Acad. Sci. U.S.A. 80, 2976-2980.

REAGAN, F. P. (1915). Vascularization phenomena in fragments ofembryonic bodies completely isolated from yolk-sac blastoderm.Anat. Rec. 9, 329-341.

RISAU, W. (1986). Developing brain produces an angiogenic factor.Proc. natn. Acad. Sci. U.S.A. 83, 3855-3859.

RISAU, W. & LEMMON, V. (1988). Changes in the vascularextracellular matrix during embryonic vasculogenesis andangiogenesis. Devi Biol. 125, 441-450.

RISAU, W., GAUTSCHI-SOVA, P. & BOHLEN, P. (1988). Endothelialcell growth factors in embryonic and adult chick brain are related

to human acidic fibroblast growth factor. EMBO J. 7, 959-962.ROMANOFF, A. F. (1960). The Avian Embryo. 1305pp. New York:

Macmillan Co.SABIN, F. (1920). Studies on the origin of blood vessels and of red

blood corpuscules as seen in the living blastoderm of chicksduring the second day of incubation. Contrib. Embryol. CarnegieInst. Washington 9, 214-262.

SARIOLA, H., EKBLOM, P., LEHTONEN, E. & SAXEN, L. (1983).Differentiation and vascularization of the metanephric kidney onthe chorioallantoic membrane. Devi Biol. 96, 427-435.

SARIOLA, H., PEAULT, B., LE DOUARIN, N., BUCK, C , DIETERLEN-LIEVRE, F. & SAXEN, L. (1984). Extracellular matrix and capillaryingrowth in interspecies chimeric kidneys. Cell Diff. 15, 43-51.

SEARLS, R. L. & JANNERS, M. Y. (1971). The initiation of limb budoutgrowth in the embryonic chick. Devi Biol. 24, 198-213.

STEWART, P. A. & WILEY, M. J. (1981). Developing nervous tissueinduces formation of blood-brain characteristics in invadingendothelial cells: A study using quail-chick transplantationchimeras. Devi Biol. 84, 183-192.

WILSON, D. J. (1983). The origin of endothelium in the developingmarginal vein of the chick wing bud. Cell Diff. 13, 63-67.

WILT, F. H. (1965). Erythropoiesis in the chick embryo: the role ofendoderm. Science 147, 1588-1590.

WINKLER, G. C. (1988). Pulmonary intravascular macrophages indomestic animal species: Review of structural and functionalproperties. Am. J. Anat. 181, 217-234.

ZACCHEI, A. M. (1961). Lo sviluppo embrionale della quagliagiapponese (Coturnix coturnix japonica). Archivio di Anat. 66,36-62.

(Accepted 5 December 1988)

relationship between vasculogenesis, angiogenesis and … · cv, cardinal vein; g, gizzard; 1,...

Documents