Abstract HNK-1 expression was studied by immuno-histochemistry in serial sections of embryonic and fetalrat hearts from 11.5 to 16.5 embryonic days. Graphic re-constructions were made to obtain detailed 3D informa-tion on the localization of immunoreactive tissues. Theantibody used appeared to stain most parts of the venoussinus and the sinuatrial transitional zone as well as theatrioventricular transitional zone, but the patterns variedthrough the different developmental stages. At 11.5 days,positive myocardium was found in the right atrium andon top of the ventricular septal primordium. At 13.5days, the left venous valve and the posterior atrial wallcontaining the orifice of the pulmonary vein were immu-noreactive, and so were the right venous valve, the sep-tum spurium and the superior, right-lateral and inferiorparts of the atrioventricular canal. From the latter, immu-noreactivity continued onto the crest of the ventricularseptum. At 15.5 days, HNK-1 positivity in the two ve-nous valves had become continuous, whereas the right-lateral part of the atrioventricular canal had lost its posi-tivity, thus making the positive areas in the superior andinferior parts of this canal discontinuous. From the ve-nous valves immunoreaction continued into the venoussinus septum but this area remained discontinuous withthe inferior part of the atrioventricular canal. It is con-cluded that the entirety of venous sinus and sinuatrialtransitional zone expresses the HNK-1 antigen and thatthe orifice of the pulmonary vein belongs to this com-plex, rather than to the embryonic atrium proper, whichis HNK-1 negative. Extrapolation of these data to the

adult human atrium leads to the conclusion hat most‘‘atrial septal structures‘‘ are of sinuatrial origin, leavingthe flap valve of the oval fossa (atrial septum primum) asthe only really atrial structure. It is suggested that theatrioventricular node is derived from the inferior portionof the atrioventricular canal, and that two expansions ofsinuatrial tissue form the substrate for anterior and poste-rior atrionodal inputs which in the literature have beendescribed as internodal tracts.

Key words Pulmonary vein · Conducting tissues ·Internodal tracts · Atrium · Venous sinus

Introduction

Several investigators (Ikeda et al. 1990; Aoyama et al.1993; Nakagawa et al. 1993; Sakai et al. 1994) have usedthe HNK-1 (Leu-7) antibody (Abo and Balch 1981) todelineate the embryonic conducting tissues. Althoughthe exact nature of the antibody is obscured by the factthat various HNK-1 antigens are expressed at differenttimes and different sites (Luider et al. 1993), the anti-body was also shown to be a good marker for the venoussinus and sinuatrial tissues (DeRuiter et al. 1995). Be-cause these tissues are known to constitute several com-ponents of the mature atrium, the marker could also helpto distinguish between these tissues and those of the em-bryonic atrium proper. To establish the 3D relationshipsof HNK-1 positive structures within the embryonic heartin more detail than had been possible before (Ikeda et al.1990; Ito et al. 1992), we have restudied the materialpreviously described by Ikeda et al. (1990). We used agraphic reconstruction method (Tinkelenberg 1979) thathas appeared to provide details comparable to thoseknown from scanning electron microscopy (Wenink andChon 1984).

We have first concentrated upon the venous part ofthe heart, to confirm that the rat pulmonary vein drainsinto the venous sinus rather than into the embryonic atri-um, as has been recently shown for the avian heart

A.C.G. Wenink (✉) · P. Symersky · M.C. DeRuiterR.E. Poelmann · A.C. Gittenberger-de GrootDepartment of Anatomy and Embryology,Leiden University Medical Center, PO Box 9602,2300 RC Leiden, The Netherlandse-mail: a.c.g.wenink@mail.medfac.leidenuniv.nlTel.: +31-71-5276682Fax: +31-71-5276680

T. IkedaDepartment of PathologyNagasaki University School of Medicine, Nagasaki, Japan

Anat Embryol (2000) 201:39–50 © Springer-Verlag 2000

O R I G I N A L A RT I C L E

Arnold C.G. Wenink · Petr SymerskyTakayoshi Ikeda · Marco C. DeRuiterRobert E. PoelmannAdriana C. Gittenberger-de Groot

HNK-1 expression patterns in the embryonic rat heartdistinguish between sinuatrial tissues and atrial myocardium

Accepted: 17 June 1999

40

(DeRuiter et al. 1995). Then, we expanded our investiga-tion to the atrioventricular area, to be able to refer to thedevelopment of the atrioventricular node and to the ques-tion of internodal pathways. The data enable us to dis-cuss the contribution of the embryonic venous sinus andthe sinuatrial transitional zone to the septal tissues of themature atrium.

Materials and methods

Wistar white rats (Kyudo, Tosu, Japan) were used. The morningwhen sperm appeared in the vaginal smears was designated as day0. Embryos and fetuses were removed at 11.5, 12.5, 13.5, 14.5,15.5 and 16.5 days. Whole embryos and fetal hearts were fixed in100% ethanol, embedded in paraffin and sectioned serially at6 µm. Deparaffinized sections were subjected to the immunostain-ing of HNK-1 by the streptavidin-peroxidase technique (Hsu et al.1981) and to hematoxylin and eosin (HE) staining. The antiseraand specific chemicals used in the procedures were as follows:normal goat serum for preabsorption of non-specific antigen,mouse IgM monoclonal anti-human Leu-7 (HNK-1) antibody(Beckton-Dickinson Immunocytometry System, Mountain View,Calif., USA) diluted 1:80 as the primary antibody, biotinylatedgoat anti-mouse IgM (Vector Laboratories, Burlingame, Calif.,USA) diluted 1:400 as the secondary antibody, peroxidase conju-gated streptavidin (BioGenex Laboratories, Dublin, Calif., USA)and 3,3′-diaminobenzidine tetrahydrochloride (0.2 mg/ml) inphosphate buffered saline (PBS) with 0.005% hydrogen peroxide(Wako Pure Chemical Industries, Osaka, Japan). After immuno-staining, the sections were counter-stained with hematoxylin. Allreactions took place at room temperature. The diluent for all anti-sera was PBS at pH 7.2. Sections of normal human lymph nodefixed in 100% ethanol served as positive control. All serial sec-tions of the heart were photographed, and these photographs wereused in the graphic reconstruction procedure (Tinkelenberg 1979;Wenink and Chon 1984). The latter method was used to depictHNK-1 immunoreactive sites individually as well as in their topo-graphical relationship with embryonic cardiac structures.

For comparison, we also studied the normal mature humanheart to describe the detailed relationships of the venous connec-tions with the atrial septum.

Results

Ideally the embryonic heart should be described as con-sisting of a set of segments and transitional (interseg-mental) zones. For the present results this would mean aclear distinction between venous sinus, sinuatrial transi-tional zone, embryonic atrium and atrioventricular tran-sitional zone. In most instances this distinction can wellbe made with the help of morphological criteria other

than immunohistochemistry. In all stages studied theatrioventricular canal (transition) is recognized as a con-stricted portion of the heart with endocardial cushionsinside. On the other hand, invagination of the venous si-nus into the atrium leads to formation of the venousvalves, which are, therefore, easily recognized as sinua-trial structures. However, in one area the criterium of in-vagination is disputable. This is the area of the pulmona-ry venous orifice, to the left of the left venous valve. Inthe following descriptions this area is referred to as the‘‘posterior atrial wall‘‘, which term is adopted from adultatrial anatomy.

At 11.5 days only two HNK-1 immunoreactive siteswere present. One was found in the ventricular septalprimordium, the other in pectinate muscles in the roof ofthe right part of the atrium. These pectinate muscles didnot yet show the immediate relationship with the sinua-trial junction as seen in later stages. At 12.5 days the on-ly difference was a slight expansion of the ventricularmyocardial positivity into the postero-inferior part of theatrioventricular canal.

In the 13.5 days embryo the HNK-1 positivity wasmuch more extensive and complex than in previous stag-es. The pectinate muscles in the right atrium werestrongly positive, and the fact that this area was relatedto an invagination of the atrial wall indicated that thiscomplex of atrial pectinate muscles belonged to what isknown as the septum spurium (Fig. 1). From there, im-munoreactivity continued into the myocardium of theright venous valve. HNK-1 expression did not uninter-ruptedly continue into the left venous valve but individu-ally this valve showed positivity as well and this contin-ued through the posterior atrial wall towards the left si-nus horn. Thus, HNK-1 immunoreactivity also surround-ed the origin of the pulmonary vein, where myocardiumas well as the mesenchymatous dorsal mesocardiumwere positive (Fig. 2). From the right venous valve,HNK-1 expression extended into the roof of the rightatrium (Fig. 3 f) from where it continued, as the septumspurium, into the right atrial anterior wall and the superi-or wall of the atrioventricular canal (Fig. 3 b, h). Theright wall of the atrioventricular canal was positive aswell and connected the areas in its superior and inferiorwalls. From the latter, positive myocardium extended onthe rim of the ventricular septal primordium (Fig. 3 b, h).

At 15.5 days the two HNK-1 positive areas of the sin-uatrial junction had become confluent (Fig. 4 f). Theproximal part of the left superior caval vein and the leftsinus horn were positive and so was the posterior atrialwall. The proximal portion of the common pulmonaryvein contained positive myocardium as well. From there,the immunoreactivity continued into the left venousvalve and surrounded the entire entrance of the right su-perior caval vein. Thus, it was also continuous with posi-tive tissue in the right venous valve and with a large areaof pectinate muscles in the roof of the right atrium andon its anterior wall. The latter area was continuous withpositive myocardium in the superior wall of the atrioven-tricular canal (Fig. 4 b, d). The right wall of the atrioven-

41

Fig. 1 A, B Rat embryo of 13.5 days. A Transverse section of theheart. Pectinate muscles in the right atrium (septum spurium) arestrongly HNK-1-positive. B Higher magnification of the boxed ar-ea in A, to show the indentation that indicates the most cranial partof the sinuatrial junction. Bars A 200 µm, B 50 µm

Fig. 2 The same embryo as in Fig. 1, to show the pulmonary vein(open arrow). The myocardium to the left of the right superiorcaval vein (S) and around the pulmonary vein express HNK-1 re-activity. Note that also the mesenchyme of the dorsal mesocardi-um (m) expresses the antigen, as well as superior (avs) and inferi-or (avi) parts of the atrioventricular canal. Bar 200 µm

42

tricular canal did not show immunoreactivity and thepositive areas in its superior and inferior walls were dis-continuous (Fig. 4 b, d). From the latter area, positivetissue coursed on the rim of the ventricular septum (bun-dle of His) and on its left side (left bundle branch; Fig. 4b, d). A further difference with the 13.5 days stage wasan extension, mainly from the right venous valve, intothe sinus septum between the entrances of the left andright sinus horns in the right atrium (Fig. 4 d).

At 16.5 days the venous sinus was still more incorpo-rated into the atrial segment. Confluent immunoreactiveareas were present around the right superior caval veinand in the two venous valves (Fig. 5 h). From there, im-munopositivity extended through the roof of the rightatrium (septum spurium) and its anterior wall into the su-perior wall of the atrioventricular canal (Fig. 5 b). Mainly

from the right venous valve, HNK-1 positive tissue wasseen to extend through the dorsal atrial wall, where it sur-rounded the orifice of the pulmonary vein, to the left su-perior caval vein (Fig. 5 d, h). Individual positivity, i.e.not continuous with either the septum spurium and thesuperior wall of th atrioventricular canal (Fig. 3 b) or thevenous sinus septum (Fig. 3 d), was seen on the crest ofthe ventricular septum and on its left side (bundle of Hisand left bundle branch). In conclusion, in the final stagewe examined there were large and confluent areas withHNK-1 positivity at the venous pole of the heart, includ-ing a small portion in the superior wall of the atrioven-tricular canal. A fully separate HNK-1 positive area wasseen in the inferior wall of the atrioventricular canal andthis continued on the ventricular septum. In contradistinc-tion to these immunoreactive structures, the developingatrial septum primum did not show any HNK-1 positivity.

A normal mature human heart was studied for compari-son. Any grooves, ridges or valves were used to demarcateits different components. Thus, we tried to make a detailedcorrelation between the morphology described above andmature atrial anatomy. The relationships of the venous con-nections with each other and with the atrial septum were asfollows. When viewed from the outside (Fig. 6 a), a largeportion of the posterior atrial wall was occupied by the en-trances of the pulmonary and caval veins. In particular, theright pulmonary veins bordered immediately upon the in-teratrial groove, while the superior and inferior caval veinshad the same relationship with the groove on the otherside. In fact, the interatrial groove seemed to separate pul-monary and systemic venous inflows. When the atrial sep-tal surface was studied from the right atrium (Fig. 6 b) sev-eral profiles appeared to be important. The entrance of thesuperior caval vein was delimited anteriorly by a muscularridge that itself formed the posterior boundary of the rightatrial appendage. This ridge came down as far as the areaof the membranous septum. The postero-inferior border ofthe entrance of the superior caval vein was formed by asimilar muscular ridge that was immediately related to theentrance of the right pulmonary veins on the other side.The anterior muscular ridge continued posteriorly into theterminal crest that descended on the posterior right atrialwall, to continue into the Eustachian valve that guarded theorifice of the inferior caval vein on its right side. Fromthere, a continuation existed as the Thebesian valve thatguarded the orifice of the coronary sinus. Between the twolatter orifices, and starting from the indentation betweenthe two valves, a ridge was present that coursed anteriorlyand ended at the membranous septum. This ridge is knownas the venous sinus septum. The posterior portion of thelimbus of the fossa ovalis corresponded with the externalinteratrial groove and appeared to be a shallow fold ratherthan an atrial septal structure.

Discussion

HNK-1 (Leu-7; Abo and Balch 1981) appears to recog-nize antigens with variable temporo-spatial expression

43

Fig. 3 a–h Graphic reconstructions of the heart of a rat embryo of13.5 days, in which the ventricular mass has been dissected away.a Frontal view. Behind the left (l) and right (r) atria, the two supe-rior caval veins (LSCV, RSCV) are visible. The atrioventricular ca-nal (av) is filled with cushion tissue. b The same, with highlight-ing of the HNK-1 positive tissues. In the roof of the right atrium,the septum spurium (sp) is positive and so is its continuation in theanterior right atrial wall, which ends in an immunoreactive area inthe superior wall of the atrioventricular canal (avs). The right wallof the atrioventricular canal is positive as well and connects thesuperior positive area with a positive area in the inferior wall ofthe canal (avi). From there, HNK-1 positivity continues on the rimof the ventricular septal primordium (H). c The same as in a, withfurther dissection of the anterior walls of the atria. In this stage,the primordium of the atrial septum (as) mainly consists of endo-cardial cushion tissue. To its left-posterior, the entrance of the pul-monary vein is arrowed. The venous sinus (vs) enters the rightatrium between the left (lv) and right (rv) venous valves. The rightvenous valve continues into the septum spurium (sp). The latterhas been cut where it courses through the anterior right atrial wall.d As c, with highlighting of the HNK-1 positive tissues. The im-munoreactivity in the right venous valve (rv) continues, throughthe septum spurium, into the anterior wall of the right atriumwhere it has been cut (asterisk). Immunoreactivity in the left ve-nous valve (lv) continues in the dorsal atrial wall, where it sur-rounds the pulmonary venous orifice, and the left sinus horn.e Posterior view, to show the left (LSCV) and right (RSCV) superi-or caval veins reaching the venous sinus just below the pulmonaryvein (pv; l left atrium, r right atrium, av atrioventricular canal).f As e, with highlighting of the HNK-1 positive tissues. Two sepa-rate areas are seen. Immunoreactivity in the right venous valve(rv) continues in the septum spurium (sp) and the superior wall ofthe atrioventricular canal (avs). Through the right wall of the atrio-ventricular canal, the latter area is continuous with the positivemyocardium on the rim of the ventricular septal primordium (H).Immunopositive tissues in the left venous valve (lv) continue inthe dorsal atrial wall, with the orifice of the pulmonary vein, andthe left sinus horn (compare Fig. 3 e). g Right lateral view inwhich, by dissection, the interior of the right atrium is shown(RSCV right superior caval vein, LSCV left superior caval vein, avatrioventricular canal filled with cushion tissue, as atrial septum,rv right venous valve). h As g, with highlighting of the HNK-1positive tissues. Immunoreactivity in the right venous valve (rv)continues in the septum spurium (sp) and the anterior wall of theright atrium, from where it reaches the superior wall of the atrio-ventricular canal (avs). The latter is, by positive myocardium inthe lateral wall of the atrioventricular canal, connected with posi-tive tissue in the inferior wall of this canal (avi). From that site,HNK-1 expression continues on the crest of the ventricular septalprimordium (H). Bar a (for a-h) 200 µm

44

patterns (Luider et al. 1993). Co-expression has beenfound with N-CAM (Kruse et al. 1984) and with the fi-bronectin receptor (Pesheva et al. 1987), and it is alsoco-localized with extracellular matrix components(Kruse et al. 1985; Hoffman and Edelman 1987; Gowdaet al. 1989). It has been specifically found to be presenton the surface of embryonic myocytes that are thought tobe involved in development of the conducting tissues(Ikeda et al. 1990; Aoyama et al. 1993; Sakai et al.1994). The present study, using a graphic reconstructiontechnique (Tinkelenberg 1979; Wenink and Chon 1984),shows the 3D relationships of HNK1-immunoreactivetissues with the embryonic cardiac structures in more de-tail than was possible in previous reports (Ikeda et al.1990; Ito et al. 1992). It shows the positive areas in suffi-cient detail to relate them specifically to the venousvalves, the septum spurium, the left and right sinus hornsand the intervening dorsal atrial wall with the pulmonaryvein. In fact, these areas are summarized as the venoussinus and the sinuatrial junction.

The pulmonary vein drains into the venous sinus

Previously, HNK-1 expression was documented in thechick and quail embryo (DeRuiter et al. 1995) and it wasshown that in certain stages the entire venous sinus waspositive, whereas in no stage did the atrial segment showany reaction. In that study, also other techniques wereused to trace development of the pulmonary vein. The

monoclonal antibody QH-1 was used to detect the first,non-luminized, strand of endothelium that is the precur-sor of the pulmonary vein and that appeared to be relatedto the venous sinus. Scanning electron microscopy wasused to show the first indication of the sinuatrial bound-aries. Thus, no doubt remains on the relationship of theprimitive pulmonary vein with the venous sinus in theavian embryo. From these methods we presently selectedthe HNK-1 immunoreaction to verify the findings in therat embryo.

As stated above, strict morphological criteria make itpossible to distinguish between venous sinus, sinuatrialtransitional zone and the embryonic atrium proper. Anexception to this is the cardiac wall, which contains theorifice of the pulmonary vein. This wall will be calledthe posterior atrial wall.

As in the bird, the HNK-1 immunopositivity is time-related in the rat embryo. Already in the 11.5 days em-bryo, part of the sinuatrial junction appeared to bestrongly positive. This early positivity concerns the pec-tinate muscles in the roof of the embryonic right atrium(Fig. 1). At 13.5 days, the venous sinus positivity andthat of the sinuatrial junction was present in two discon-tinuous areas. One comprised the left sinus horn and theleft venous valve and included the pulmonary vein, theother comprised the roof of the right atrium and the rightvenous valve (Fig. 3). In the 15.5 and 16.5 days embry-os, immunoreactivity was seen around all venous con-nections including that of the pulmonary vein, and in themyocardium of the posterior atrial wall, both venousvalves and the roof of the right atrium (Figs. 4, 5). Thesedata suggest that the entire venous sinus and sinuatrialjunction react with the HNK-1 antibody, and that the pul-monary vein belongs to this segment of the embryonicheart rather than to the atrial segment. This conclusion isconsistent with what has been found recently in themouse embryo between days 9 and 10 (Tasaka et al.1995), although contrary to the opinion of Webb et al.(1996), who found in mouse embryos of 17 and 18 daysthat the pulmonary vein is too much to the left of the leftvenous valve to belong to the venous sinus. Also inyounger stages these latter authors (Webb et al. 1998)have found the precursor of the pulmonary vein to bediscrete from the venous sinus.

Our findings are similar to those of Nakagawa et al.(1993), including the HNK-1 reactivity on both sides ofthe pulmonary vein. These authors describe two separatereactive areas behind the dorsal atrial wall, as we de-scribe in the 13.5-day embryo. Their oldest embryo stud-ied, however, was 14.5 days old, whereas we show thatat 15.5 and 16.5 days the two positive areas are conflu-ent.

The sinuatrial transitional zone

When all HNK-1 positive tissues at the venous pole ofthe heart are part of the venous sinus or the sinuatrialtransitional zone, particular attention has to be paid to

45

Fig. 4 a–f Graphic reconstructions of the heart of a 15.5-day ratembryo. a Frontal view to show the left (LSCV) and right (RSCV)superior caval veins, the left (l) and right (r) atria, and the atrio-ventricular canal which is divided in a mitral (m) and a tricuspid(t) orifice. b The same, with highlighting of the HNK-1 positivetissues. Immunoreactive tissue around the right superior caval vein(rs) continues into the right venous valve and the venous sinusseptum (ss). The positivity in the right venous valve is also contin-uous (connection not visible) with the septum spurium (sp) andthe superior wall of the atrioventricular canal (avs). The positivetissue in the right venous valve is also continuous with the dorsalatrial wall (which contains the orifice of the pulmonary vein) andthe tissue around the left superior caval vein (ls). Separate immu-noreactivity is seen in the inferior wall of the atrioventricular ca-nal (avi), which continues in the bundle of His (H) and the leftbundle branch (lb). c Right lateral view (LSCV left superior cavalvein, RSCV right superior caval vein, l left atrium, r right atrium,m mitral orifice, t tricuspid orifice). d As c, with highlighting ofthe immunoreactive areas. This view shows that there is no con-nection between either the superior wall of the atrioventricular ca-nal (avs) or the venous sinus septum (ss) and the inferior wall ofthe atrioventricular canal (avi). The latter continues into the Hisbundle (H) and left bundle branch (lb). e Posterior view, showingthe two superior caval veins (LSCV, RSCV) entering the heart justbelow the pulmonary vein (pv; l left atrium, r right atrium). f As e,with highlighting of the immunoreactive tissues. HNK-1 positivityaround the left superior caval vein (ls) continues in the dorsal atri-al wall (with the orifice of the pulmonary vein) and the left venousvalve (lv). Positivity in the latter is also continuous with the rightvenous valve (rv). From there, HNK-1 reactivity continues (con-nection not visible) in the septum spurium (sp) and the superiorwall of the atrioventricular canal (avs). More inferiorly, positivityof the right venous valve continues in the venous sinus septum (ss;H His bundle, lb left bundle branch). Bar a (for a–f) 200 µm

46

the strongly positive cells in the pectinate muscles of theright atrial roof. This area was previously identified asthe septum spurium, i.e. the cranial confluence of thetwo venous valves (Ikeda et al. 1990; Nakagawa et al.1993). Indeed, this area is directly continuous with themusculature of the right venous valve (Fig. 3 d) and laterthat of both venous valves (Figs. 4 b, 5 b). The deep in-dentation of the atrial wall at this site (Fig. 2) helps itsidentification as an intersegmental transition. The anteri-or extension of this area becomes then of considerableinterest because it is continuous with HNK-1 positivecells in the superior wall of the atrioventricular canal(Figs. 3 d, 4 b, 5 d). In previous descriptions of the 3Drelationships of the intersegmental specialized rings ortransitional zones (Wenink 1976; Gittenberger-de Grootet al. 1995), the inner curvature of the looped embryonicheart was described as the site where all transitionalzones are in contact. As far as the sinuatrial transition isconcerned, this was shown to reach the inner curvaturefrom posterior and inferior. The present data on the sep-tum spurium, however, add an antero-superior route to

the contacts between transitional zones. Asami andKoizumi (1995) have alluded to the changing relation-ships of the sinuatrial transition in the human heart bysuggesting that the atrial roof enlarges and elevates dur-ing development. This would explain a relatively longanterior route in later stages, which would find its basisin rather short distances in earlier phases of morphogen-esis. Others have simply termed this anterior route an‘‘internodal pathway‘‘ (Ikeda et al. 1990; Nakagawa etal. 1993), which denomination relates HNK-1 immuno-reactivity to the conducting tissues.

The precursors of the conducting tissues

We have no doubt about the HNK-1 positivity of the en-tire venous sinus, but it should be noted that there arevarious HNK-1 antigens with different spatio-temporalexpression patterns (Luider et al. 1993). This is consis-tent with the present description of immunoreactivity inboth myocardial and mesenchymatous tissues (Fig. 2).The HNK-1 epitope seems to be related with cell-cell orcell-substrate interactions (Sakai et al. 1994). The HNK-1 immunoreactivity may also be brought into connectionwith the relatively high glycogen content of the precur-sors of the conducting tissues (Virágh and Challice1977a, b). Knaapen et al. (1997) have suggested that theenzyme glycogen synthase kinase-3, which regulatesglycogen synthesis (Plyte et al. 1992) but which is alsoinvolved in the regulation of nerve growth factor recep-tor proteins (Taniuchi et al. 1986; Guan et al. 1991) playsa role in the co-existence of HNK-1 positivity and highglycogen content. Neurotrophin-3 mRNAs have beenspecifically detected in Leu-7 positive myocytes of theembryonic rat heart (Hiltunen et al. 1996). As is the casewith the high glycogen content, in many reports theHNK-1 (or Leu-7) immunoreactivity has been describedas a marker for the specialized conducting tissues (Ikedaet al. 1990; Aoyama et al. 1993; Nakagawa et al. 1993;Sakai et al. 1994). From the present study it is evidentthat the HNK-1 antibody stains more than these tissues.For instance, the immunoreactivity that surrounds the su-perior caval vein (Figs. 4 b, 5 d) seems to be more exten-sive than can be accounted for by the sinus node. Still,there are reasons to consider a special functional statusfor the myocardium of the venous sinus in the heart.Special attention was recently given to myocytes (i.e. notsmooth muscle cells) that form the wall of the superiorcaval vein in the rat embryo (Endo et al. 1996), and inadult rats node-like cells have been described to envelopthe pulmonary vein and a potential pacemaker activitywas claimed at this site (Masani 1986).

Probably, not all HNK-1-positive tissues will give riseto a specific part of the mature conducting system, butthe reverse seems nevertheless to be true: all parts of thefuture conducting system are HNK-1-positive in certaindevelopmental stages. A similar problem existed in aprevious study on development of the human conductingsystem (Wenink 1976) that suggested that all parts of the

47

Fig. 5 a–h Graphic reconstructions of the heart of a 16.5-day ratembryo. a Right lateral view, showing the two superior caval veins(LSCV, RSCV), both atria (l, r) and the atrioventricular canal,which is divided into a mitral (m) and a triscuspid (t) orifice.b The same, with highlighting of the HNK-1-positive tissues. Pos-itivity around the right superior caval vein (rs) is seen to continueinto the septum spurium (sp) and the superior wall of the atrioven-tricular canal (avs), and also into the venous sinus septum (ss). Itis also continuous with positive tissue in the dorsal atrial wall andaround the left superior caval vein (ls). This complex system ofpositive tissues has no connection with the HNK-1 reactive myo-cardium in the inferior wall of the atrioventricular canal (avi) andthe bundle of His (H). c As a, after dissection of the lateral wall ofthe right atrium and the septum spurium (sp), where it takes offfrom the right venous valve (rv). Between the two entrances of thesinus horns, the venous sinus septum (ss) is seen to course antero-superiorly. d As c, with highlighting of the immunoreactive tissue.Because the septum spurium has been taken away (sp) from theright venous valve (rv), the positive tissue in the left venous valve(lv) can be seen to continue in the posterior atrial wall (with alarge opening for the orifice of the pulmonary vein) and in thewall of the left superior caval vein (ls; ss venous sinus septum). eAnterior view of the dissected atria, to show the entrance of theright superior caval vein (RSCV) between the left (lv) and right(rv) venous valves, and that of the left superior caval vein (LSCV)between the venous sinus septum (ss) and the right venous valve.Behind the atrial septum (as) the entrance of the pulmonary vein isvisible (sp take off of the septum spurium – which has been takenaway). f As e, with highlighting of the HNK-1-positive tissues.Immunoreactivity around both caval veins, in the dorsal atrial walland in the venous sinus septum (ss) are all continous. Positive tis-sue in the septum spurium (sp) has been dissected at its take-offfrom the right venous valve. g Posterior view. The superior cavalveins (LSCV, RSCV) enter the heart just below the pulmonary vein(pv; l left atrium, r right atrium). h As g, with highlighting of theentire complex of immunoreactive tissues. HNK-1 positivity in thedorsal atrial wall (with the orifice of the pulmonary vein) is con-tinuous with positivity around the left superior caval vein and inthe left venous valve (lv) as well as the right venous valve (rv).From the latter, immunoreactivity continues (connection not visi-ble) in the septum spurium (sp) and the superior wall of the atrio-ventricular canal (avs). Continuity also exists with positive tissuesin the venous sinus septum (ss). This complex is discontinuouswith positivity in the His bundle (H). Bar a (for a–h) 200 µm

final conducting tissues are derived from a set of inter-segmental myocardial rings. That study, based on routinehistology, also suggested that some parts of the conduct-ing tissues as found in congenital malformations wouldbe derived from parts of the ring system that can no lon-ger be traced in the normal adult heart. Similarly, routinehistology of human fetal hearts formed the basis ofgraphic reconstructions of ‘‘pale cells‘‘ in the atrium thatconstituted a sinuatrial ring in these stages (Gittenberger-de Groot and Wenink 1978), and which coincided withinternodal pathways as described by others (James 1963;Netter 1969).

It is important to note that an initial continuity be-tween the sinuatrial and atrioventricular nodal areas ispresent in the roof of the right atrium, whereas in thebottom of the atrium these tissues are discontinuous (Fig.3 h). The complete continuity is temporary. At 15.5 daysthe right portion of the atrioventricular ring is no longerHNK-1 positive. Figure 5 b shows these relationships inthe 16.5-day embryo, in which immunoreactivity in thevenous valves continues through the roof and anteriorwall of the right atrium, i.e. the septum spurium, andends in the superior part of the atrioventricular ring. The

positive area in this ring is discontinuous with a similararea in its inferior part, making this and its continuationin the primary ring on the ventricular septum, i.e. the Hisbundle, a separate site of HNK-1 expression. Incorpora-tion of the venous sinus has progressed considerably, butthe HNK-1 expression in the venous sinus septum doesnot contact the atrioventricular junction (Fig. 5 b, d).

HNK-1 expression in two separate parts of the atrio-ventricular ring has led to the description of two atrioven-tricular nodes (Ikeda et al. 1990). We are unable to con-firm the fusion of these two primordia that was reported totake place in the rat fetus of 18 days (Aoyama et al. 1993),but still an alternative hypothesis may be offered. It maywell be that the immunoreactive area in the inferior por-tion of the atrioventricular canal is the sole precursor ofthe atrioventricular node. The HNK-1 expressing myocar-dium in the superior wall of the canal would then be iden-tical to the atrioventricular ring tissue that was describedto produce the anomalous anterior nodes in congenitallymalformed hearts (Wenink 1976, 1978, 1979). In bis-di-amine treated rats the posterior atrioventricular node hasbeen observed to be poorly developed while the anteriornode showed intense HNK-1 expression (Ito et al. 1992).

Our view that the inferior part of the atrioventricularring gives rise to the atrioventricular node is supportedby other evidence on specialization of the atrioventricu-lar ring (Anderson et al. 1974; Wessels et al. 1992). Theproximity of sinuatrial tissues to this area would then notcontribute to the atrioventricular node itself, but theywould provide the atrial inputs to the node. In the rabbitheart the latter have been termed the “open node” as op-posed to the more anterior “enclosed node”, which is en-cased in fibrous tissue and continues directly into the Hisbundle (Tranum-Jensen 1976). A similar concept ofatrioventricular node morphology has been described inthe dog (Racker 1989). In this animal the node was re-ported to find itself encased in fibrous tissue, like the hu-

48

Fig. 6 a, b The venous part of a human heart (adult male).a Posterior view of the atria, to show the close proximity of theright pulmonary veins (rp) and the superior (SCV) and inferior(ICV) caval veins, which are only separated by a shallow interat-rial groove. b The interior of the right atrium, to show the orific-es of the superior (SCV) and inferior (ICV) caval veins and thatof the coronary sinus (sc). Anteriorly, a muscular bar (m) delim-its the medial atrial wall (MAW), which is the medial wall of theright atrial appendage. A continuous circle is formed by thismuscular bar, the terminal crest (tc), the Eustachian valve (ev)and the sinus septum (ss). The circle, which marks the sinuatrialjunction, is closed at the membranous septum (ms). The posteri-or limbus (l) of the oval fossa (OF) is a shallow fold which cor-responds with the interatrial groove seen in 6a (TRIC tricuspidanulus, rp right pulmonary veins)

man ‘‘Knoten‘‘ in the original description by Tawara(1905), whereas nodal inputs are provied by a proximalatrioventricular bundle (PAVB) and a superior atrionodalbundle (SAB). These nodal inputs could well be of sinu-atrial origin. The SAB of Racker (1989) seems to becomparable to the continuation of the septum spuriuminto the superior wall of the atrioventricular canal in ourmaterial (Fig. 3 h), and her PAVB must be related to thevenous sinus septum (Fig. 3 d). Indeed, Racker (1989)concludes that the sinuventricular conducting systemconsists of sinuatrial and atrioventricular components.

The presence of internodal tracts (James 1963, 1985;Liebman 1985) has been denied by several authors (Janseand Anderson 1974; Becker and Anderson 1976; Ander-son and Becker 1985), but all parts of the internodal andinteratrial pathways as described by James (1963) showtemporary HNK-1 expression (Fig. 5 b, h). The presentstudy shows that any connection, be it “specialized” ornot, between the sinus node and the atrioventricular nodemust by necessity be derived from the sinuatrial junctionand not from the primitive atrial tissues. Such connectionswill certainly join the node at the proximal atrioventricularbundle and the superior atrionodal bundle of Racker(1989) and are comparable to the anterior and posterior in-ternodal tracts (James 1963). They represent the dual in-puts into the atrioventricular node which by a fine tuningof processes of cancellation and summation of wave frontsare responsible for atrioventricular conduction (Meijlerand Janse 1988). The present view is also in agreementwith the findings of anatomically as well as electrophysio-logically distinct cells that compose a major anterior inputto the atrioventricular node (Antz et al. 1997).

Atrial septal structures in the mature heart

Our search for the distinction between venous sinus andsinuatrial transition on one hand and the atrial segmenton the other is easily extrapolated to the mature heart. Asshown in Fig. 6 b, the wall of the superior caval venousentrance comprised between the anterior and posteriormuscular ridges clearly belongs to the embryonic venoussinus. The anterior ridge must then be sinuatrial, and thefact that it reaches down to the membranous septum, i.e.the atrioventricular region, fits with our discussion onthe anterior nodal inputs (“anterior internodal tract”).Only the myocardium anterior to this ridge is of embry-onic atrial nature. But this wall does not belong to the“atrial septum”, as Racker (1989) has rightly observed.She has termed this wall, which belongs to the right atri-al appendage, the medial atrial wall (MAW) and she de-scribes the base of the MAW as separating the interven-tricular septum from the actual atrial septum, which thenbecomes limited to a small oval area within the rightatrium. The terminal crest, which connects the entranceof the superior caval vein with the valve of the inferiorcaval vein, is obviously a remnant of the right venousvalve and is a sinuatrial structure. It corresponds with theposterior nodal inputs (“posterior internodal tract”). The

smooth right atrial wall to the left of this crest must thenbe derived from the embryonic venous sinus. This area islimited by the limbus of the fossa ovalis, which corre-sponds to the external intratrial groove as seen in Fig. 6a. This portion of the limbus is no more than a shallowfold and can hardly be called an atrial septal structure.The postero-inferior part of the mature atrial septum,which borders upon the anulus of the tricuspid valve, hasbeen recognized as part of the embryonic venous sinuspreviously (Anderson et al. 1976). Comparison of thepresent Fig. 5 c and d with Fig. 6 b confirms that also thetissues inferior to the oval fossa stem from the venous si-nus. Particularly the venous sinus septum (Fig. 5 c) is alarge structure that courses antero-superiorly to the leftof the tricuspid orifice. We have to conclude that the on-ly real interatrial structure that is left is the valve of theforamen ovale, i.e. the embryonic atrial septum primum.

Conclusions

In summary, the present study shows that all parts of thevenous sinus and sinuatrial transition are HNK-1-immu-noreactive in some stage of development, and that thepulmonary vein originally drains into the venous sinus. Italso indicates that most of the structures that are usuallydescribed as part of the atrial septum in reality are part ofthe venous sinus and the sinuatrial junction. Finally, thetissues that, in the right atrium, constitute the inputs intothe atrioventricular node share their derivation from thesinuatrial junction and are developmentally distinct fromatrial working myocardium.

References

Abo T, Balch CM (1981) A differentiation antigen of human NKand K cells identified by a monoclonal antibody (HNK-1). JImmunol 17:1024–1029

Anderson RH, Becker AE (1985) Pathways of preferential atrialconduction. Letter. Br Heart J 53:350–351

Anderson RH, Davies MJ, Becker AE (1974) Atrioventricular ringspecialized tissue in the normal heart. Cardiology 2:219–230

Anderson RH, Becker AE, Wenink ACG, Janse MJ (1976) Thedevelopment of the cardiac specialized tissue. In: Wellens HJJ,Lie KI, Janse MJ (eds) The conduction system of the heart.Structure, function and clinical implications. Stenfert Kroese,Leiden, pp 3–28

Antz M, Scherlag BJ, Patterson E, Otomo K, Tondo C, Pitha J,Gonzalez MD, Jackman WM, Lazzara R (1997) Electrophysi-ology of the right anterior approach to the atrioventricularnode: studies in vivo and in the isolated perfused dog heart. JCardiovasc Electrophysiol 8:47–61

Aoyama N, Kikawada R, Yamashina S (1993) Immunohistochemi-cal study on the development of the rat heart conduction systemusing anti-Leu-7 antibody. Arch Histol Cytol 56:303–315

Asami I, Koizuki K (1995) Development of the atrial septal com-plex in the human heart: contribution of the spina vestibuli. In:Clark EB, Markwald RR, Takao A (eds) Developmental mech-anisms of heart disease. Futura, Armonk, NY, pp 255–260

Becker AE, Anderson RH (1976) Morphology of the human atrio-ventricular junctional area. In: Wellens HJJ, Lie KI, Janse MJ(eds) The conduction system of the heart. Structure, functionand clinical implications. Stenfert Kroese, Leiden pp 263–286

49

DeRuiter MC, Gittenberger-de Groot AC, Wenink ACG, Poel-mann RE, Mentink MMT (1995) In normal development pul-monary veins are connected to the sinus venosus segment inthe left atrium. Anat Rec 243:84–92

Endo H, Ogawa K, Kurohmaru M, Hayashi Y (1996) Develop-ment of cardiac musculature in the cranial vena cava of ratembryos. Anat Embryol 193:501–504

Gittenberger-de Groot AC, Wenink ACG (1978) The specializedmyocardium in the fetal heart. In: VanMierop LHS, Oppen-heimer-Dekker A, Bruins C (eds) Embryology and teratologyof the heart and great arteries. Boerhaave Series for Postgradu-ate Medical Education, vol 13, Leiden University Press, TheHague, pp 15–24

Gittenberger-de Groot AC, Bartelings MM, Poelmann RE (1995)Overview: cardiac morphogenesis. In: Clark EB, MarkwaldRR, Takao A (eds) Developmental mechanisms of heart dis-ease. Futura, Armonk NY, pp 157–168

Gowda DC, Margolis RU, Margolis RK (1989) Presence of theHNK-1 epitope on poly(N-acetyl lactosaminyl) oligosaccharideand identification of multiple core proteins in the chondroitinsulfate proteoglycans of brain. Biochemistry 28:4468–4474

Guan RJ, Khatra BS, Cohlberg JA (1991) Phosphorilation of bo-vine neurofilament proteins by protein kinase Fa (glycogensynthase kinase-3). J Biol Chem 266:8262–8267

Hiltunen JA, Arumäe U, Moshnyakov M, Saarma M (1996) Ex-pression of mRNAs for neurotrophins and their receptors indeveloping rat heart. Circ Res 79:930–939

Hoffman S, Edelman GM (1987) A proteoglycan with HNK-1 an-tigenic determinants is a neuron-associated ligand for cytotac-tin. Proc Natl Acad Sci USA 84:2523–2527

Hsu SM, Raine L, Fanger H (1981) Use of avidin-biotin-peroxi-dase complex (ABC) in immunoperoxidase techniques: a com-parison between ABC and unlabeled antibody (PAP) proce-dures. J Histochem Cytochem 29:577–580

Ikeda T, Iwasaki K, Shimokawa I, Sakai H, Ito H, Matsuo T(1990) Leu-7 immunoreactivity in human and rat embryonichearts, with special reference to the devepment of the conduc-tion tissue. Anat Embryol 182:553–562

Ito H, Iwasaki K, Ikeda T, Sakai H, Shimokawa I, Matsuo T(1992) HNK-1 expression pattern in normal and bis-diamineinduced malformed developing rat heart: three dimensional re-construction analysis using computer graphics. Anat Embryol186:327–334

James TN (1963) The connecting pathways between the sinusnode and AV-node and between the right and the left atrium inthe human heart. Am Heart J 66:498–508

James TN (1985) Pathways of preferential atrial conduction. Re-ply. Br Heart J 53:351

Janse MJ, Anderson RH (1974) Specialized internodal atrial path-ways – fact or fiction? Eur J Cardiol 2:117–136

Knaapen MWM, Vrolijk BCM, Wenink ACG (1997) Ultrastruc-tural changes of the myocardium in the embryonic rat heart.Anat Rec 248:233–241

Kruse J Mailhammer R, Wernecke H, Faissner A, Sommer I, Gori-dis C, Schachner M (1984) Neural cell adhesion moleculesand myelin-associated glycoprotein share a common carbohy-drate moiety recognized by monoclonal antibodies L2 andHNK-1. Nature 311:153–155

Kruse J, Keilhauer G, Faissner A, Timpl R, Schachner M (1985)The J1 glycoprotein: a novel nervous system cell adhesionmolecule of the L2/HNK-1 family. Nature 316:146–148

Liebman J (1985) Are there internodal tracts? Yes. Int J Cardiol7:174–185

Luider TM, Bravenboer N, Meijers C, Van der Kamp AWM,Tibboel D, Poelmann RE (1993) The distribution and charac-terization of HNK-1 antigens in the developing avian heart.Anat Embryol 188:307–316

Masani F (1986) Node-like cells in the myocardial layer of thepulmonary vein of rats: an ultrastructural study. J Anat145:133–142

Meijler FL, Janse MJ (1988) Morphology and electrophysiologyof the mammalian atrioventricular node. Physiol Rev68:608–647

Nakagawa M, Thompson RP, Terracio L, Borg TK (1993) Devel-opmental anatomy of HNK-1 immunoreactivity in the embry-onic rat heart: co-distribution with early conduction tissue.Anat Embryol 187:445–460

Netter FH (1969) The heart. The Ciba collection of medical illus-trations, vol 5. Ciba Pharmaceutical, Summit NJ, p 13

Pesheva P, Horwitz AF, Schachner M (1987) Integrin, the cell sur-face receptor for fibronectin and laminin, expresses theL2/HNK-1 and L3 carbohydrate structures shared by adhesionmolecules. Neurosci Lett 83:303–306

Plyte SE, Hughes K, Nikolakaki E, Pulverer BJ, Woodgett JR(1992) Glycogen synthase kinase-3: function in oncogenesisand development. Biochim Biophys Acta 1114:147–162

Racker DK (1989) Atrioventricular node and input pathways: acorrelated gross anatomical and histological study of the ca-nine atrioventricular junctional region. Anat Rec 224:336–354

Sakai H, Ikeda T, Ito H, Nakamura T, Shimokawa I, Matsuo T(1994) Immunoelectron microscopic localization of HNK-1 inthe embryonic rat heart. Anat Embryol 190:13–20

Taniuchi M, Johnson EM, Roach PJ, Lawrence JC (1986) Phos-phorylation of nerve growth factor receptor proteins in sym-pathetic neurons and PC12 cells. J Biol Chem 261:13342–13349

Tasaka H, Krug EL, Markwald RR (1995) Origin of the orifice ofthe pulmonary vein in the mouse. In: Clark EB, Markwald RR,Takao A (eds) Developmental mechanisms of heart disease.Futura, Armonk, NY, pp 347–350

Tawara S (1905) Die Topographie und Histologie der Brückenfa-sern. Ein Beitrag zur Lehre von der Bedeutung der Purkinjes-chen Faden. Zentralbl Physiol 19:70–76

Tinkelenberg J (1979) Graphic reconstruction, micro-anatomywith a pencil. J Audiovis Media Med 2:202–106

Tranum-Jensen J (1976) The fine structure of the atrial and atrio-ventricular (AV) junctional specialized tissues of the rabbitheart. In: Wellens HJJ, Lie KI, Janse MJ (eds) The conductionsystem of the heart. Structure, function and clinical implica-tions. Stenfert Kroese, Leiden, pp 55–84

Virágh S, Challice CE (1977 a) The development of the conduc-tion system in the mouse embryo heart. I. The first embryonicconduction pathway. Dev Biol 56:382–396

Virágh S, Challice CE (1977 b) The development of the conduc-tion system in the mouse embryo heart. II Histogenesis of theatrioventricular node and bundle. Dev Biol 56:397–411

Webb S, Brown NA, Anderson RH (1996) The structure of themouse heart in late fetal stages. Anat Embryol 194:37–47

Webb S, Brown NA, Wessels A, Anderson RH (1998) Develop-ment of the murine pulmonary vein and its relationship to theembryonic venous sinus. Anat Rec 250:325–334

Wenink ACG (1976) Development of the human cardiac conduct-ing system. J Anat 121:617–631

Wenink ACG (1978) The conducting tissues in primitive ventriclewith outlet chamber. Two different possibilities. J Thorac Car-diovasc Surg 75:747–753

Wenink ACG (1979) Congenitally complete heart block with aninterrupted Mönckeberg sling. Eur J Cardiol 9:89–99

Wenink ACG, Chon Y (1984) The value of graphic reconstruc-tions. Comparison with scanning electron microscopy. AnatRec 210:537–540

Wessels A, Vermeulen JLM, Verbeek FJ, Virágh S, Kalman F, La-mers WH, Moorman AFM (1992) Spatial distribution of “tis-sue specific” antigens in the developing human heart and skel-etal muscle. III. An immunohistochemical analysis of the dis-tribution of the neural tissue antigen G1N2 in the embryonicheart: implications for the development of the atrioventricularconduction system. Anat Rec 232:97–111

50