Cell Tiss. Res. 167, 373-385 (1976)
Cell and Tissue Research 9 by Springer-Verlag 1976
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments A Scanning and Transmission Electron Microscopic Study* W. Vogel, V. Vogel, and M. Pfautsch Institutes of Anatomy and of Pathology, University of Tiibingen and Institute for Medical Physics, University of MOnster, Germany
Summary. The origin of arterio-venous anastomoses, connecting the efferent filament artery (EFA) with the central venous sinus (CVS) o f gill filaments can be well discerned by scanning electron microscopy in the rainbow trout. Corresponding vessels between the afferent filament artery and the CVS could not be detected with the techniques applied. AVA-specific endothelial cells are characterized by their bulky shape and their microvillous surface. The general m o r p h o l o g y o f A V A ' s in Salmo gairdneri is very similar to that of A V A ' s in Tilapia mossambica (Vogel et al., 1974) but they are much longer in the trout. N o filament whorls have been encountered in A V A endothelia of Salmo gairdneri. Key words: Gills - Salmo gairdneri - Specialized endothelia venous anastomoses - Electron microscopy.
Arterio-
Introduction The general arrangement and ultrastructure of arterio-venous anastomoses (AVA's) in gill filaments o f a cichlid fish, Tilapia mossambica, has been described recently (Vogel et al., 1973, 1974) 1. One characteristic feature of these A V A ' s in T. mossambica are highly specialized endothelial cells with numerous tentacular microvilli at their surface. These AVAospecific cells usually protrude into the lumen o f the corresponding gill filament artery at the anastomosis' origin. It is therefore possible to demonstrate the origin of A V A ' s in longitudinally sliced efferent gill filament arteries in T. mossambica by scanning electron microscopy (unpublished observations).
Send offprint requests to: Dr. Walter O.P. Vogel, Anatomisches Institut der Universitfit Tiibingen, D-7400 T/ibingen, Osterbergstr. 3, Federal Republic of Germany * This study is dedicated to Prof. Dr. W. Graumann, Director of the Institute of Anatomy, University of Tiibingen, on the occasion of his 60th birthday. It was supported by the Deutsche Forschungsgemeinschaft 1 In this species most of the AVA's are located between the efferent filament artery and a central venous sinus (CVS) while those between the afferent filament artery and the CVS are extremely rare
374
W. Vogel et al.
~E
b|e--~L~ /
fe
It
pc
CVS
\|174 e~'
a[a .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
~A~
Fig.
1. Diagram of cross sectioned gill filament. Sections for T E M were cut in this plane. The plane of sectioning for the SEM preparations was perpendicular to it where indicated by broken lines at the afferent (A) and at the efferent (E) side of the filament. Thick arrows: directions of scanning. Thin arrows: Main blood flow direction in the respiratory lamellae. 9 usual site of AVA origin in the wall of the efferent filament artery (EFA). ** region of the central venous sinus (CVS) where all of the A V A ' s investigated in this respect opened into the CVS. AFA afferent filament artery, ala afferent lamellar arteriole, ble basal lamellar endothelial cells, ela efferent lamellar arteriole, fc filament cartilage, fe filament epithelium, ll lamellar lacuna (lamellar blood space), pe pillar cells
In spite of the pioneer work of Steen and Kruysse (1964) in this field it is still controversial, whether similar vascular connections do exist in gill filaments of other teleosts. Investigations carried out mainly on salmonids (Steen and Kruysse, 1964; Newstead, 1967; Richards and Fromm, 1969; Morgan and Tovell, 1973; Morgan, 1974; Cameron, 1974; cf. also Hughes and Morgan, 1973; Cameron and Polhemus, 1974) have yielded contradictory results. As such AVA's in fish gill filaments may be important for respiratory and for
Fig. 2. Efferent side of gill filament sectioned longitudinally: Luminal view of efferent filament artery (EFA) and of three efferent lamellar arterioles (ela). Two of the latter share a c o m m o n outflow to the EFA. The origin of two arterio-venous anastomoses is well recognizable by the A V A specific endothelial cells (se) bulging between the endothelial cells of the artery. A few of the recessus of the central venous sinus (CVS) reaching far around the arterial vessels still contain some red blood cells (rbc). Arrows indicate blood flow direction (630: 1)
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
375
Fig. 3. Closer view of an AVA-origin in the wall of the efferent filament artery, a e arterial endothelial cell, s e AVA-specific endothelia with bulky cell projections (arrows), l lumen of the AVA (2,750: 1)
376
W. Vogel et al.
osmoregulatory gill functions, we decided to search again for these anastomoses in gill filaments of Salmo gairdneri, applying mainly scanning electronmicroscopic techniques. Material and Methods The fish, Salmo gairdneri (Richardson), were purchased from a local dealer and kept in well oxygenated running tap water (14~ for up to 14 h prior to use. 3 fish, 25-35 cm long, weighing 310450 g, were studied. Each one was anaesthetized with MS 222 (0.5 g/l) for 2 min. The vascular system of the fish was perfused through the ventral aorta with the following perfusion mixtures: Ringer solution (pH 7.3; osmolality: 320 mOsm/Kg H20 ) for 0.5 min, followed immediately by a 2% glutaraldehyde solution buffered with Na-cacodylate (pH 7.3; osmolality: 430 mOsm/Kg H20). The perfusion pressure was adjusted to 40q50 Torr. After 10 min of perfusion, gill filaments from the middle of the first gill arches of both sides were excised and immersed in a fresh portion of the same fixative for 3-4 h at 4~ The tissue was then rinsed in cold cacodylate buffer supplemented with saccharose (osmolality: 320 mOsm/Kg H20), postfixed in 2% OsO4 adjusted with Na-cacodylate buffer to pH 7.3, for 2 h and partially dehydrated in a graded ethanol series in steps of 10% starting with 30% ethanol. In 70% ethanol the material was divided into 2 groups:
1. Scanning Electron Microscopy ( S E M )
Immersed in a fresh cold ethanol solution (70%) the efferent or afferent side of single gill filaments was cut longitudinally by means of a razor blade. Under the control of a dissecting microscope this knife was adjusted with a micromanipulator such, that sectioning took place through the middle of either the afferent or efferent filament artery (cf. Fig. 1). Thus fairly long pieces of these vessels could be opened. The specimens were then processed further for drying above the critical point of Freon 13 (Fromme et al., 1972). Dried specimens were mounted on aluminium specimen studs with conducting silver paint and coated with gold by means of a Leitz sputter coating unit. They were examined in a scanning electron microscope "Stereoscan MK I " (Cambridge Instruments Ltd.).
2. Transm&sion Electron Microscopy ( T E M )
Whole filaments were further dehydrated through graded ethanol to propylene oxide and embedded in Araldite. Semi- and ultrathin sections were cut with a Reichert OmU3 ultramicrotome. The plane of sectioning is shown in Figure 1. Semithin sections were stained with toluidine blue for light microscopy. Ultrathin sections stained with uranyl-acetate (Watson, 1958) and lead citrate (Venable and Goggeshall, 1965) were examined with a Siemens Elmiskop 102.
Fig. 4. Oblique section of an AVA at its origin in the efferent filament artery (EFA). AVA-specific endothelia (se) bear microvilli as well as irregular, bulky protrusions occasionally with ruptures of the cell membrane (arrow). ae arterial endothelial cells, sm smooth muscle cells, gc bright cell with electron dense granules of various size, cc cover cells of the AVA in immediate contact with smooth muscle cells of the EFA. ve venous endothelial cell. Parts of the central venous sinus (CFS) are very often in such a close vicinity to the EFA. No AVA, however, has been observed to open immediately into the CVS in this area (6,000: 1) Fig. 5. Origin of an AVA as seen from the lumen of the efferent filament artery, ae arterial endothelial cell, se AVA-specific endothelial cell, arrows: microvilli at the border of normal endothelial cells, p bulky protrusion on the surface of AVA endothelial cell (6,870 : 1)
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
377
378
W. Vogel et al.
Results
The present investigation confirms that arterio-venous anastomoses (AVA's) do exist between the efferent filament artery (EFA) and a sinusoidal central venous system (CVS) of S. gairdneri gill filaments. Most of these AVA's are located in that part of the EFA adjacent to the filament core, where they are easily accessible for SEM examination (cf. Fig. 1). In filamental cross sections, some AVA's could also be seen to originate from efferent lamellar arterioles. Judging from the SEM preparations the AVA's appear at irregular intervals along the wall of the EFA. They seem to be less numerous than the lamellar pairs in the filaments. The microvilli at the surface of AVA-specific endothelial cells, which protrude into the lumen of efferent filament arteries, greatly facilitate the location of AVA's (i.e. their origin) by scanning electron microscopy (Fig. 2). In afferent filament arteries, however, no similar cells of corresponding AVA's have been detected with this technique (Fig. 10). In a few filaments afferent lamellar arterioles have also been investigated, but again no AVA's could be identified here. In the EFA, microvilli are not entirely confined to AVA endothelial cells (cf. Fig. 5), but here they are definitely concentrated and usually longer than in normal endothelia. These microvilli are apparently not evenly distributed over the luminal surface of the AVA cells, but are preferably located around the AVA lumen and also in the area of contact with normal endothelia. In addition to the microvilli, a few, rather plump protrusions can be seen in SEM pictures of AVA endothelia (Figs. 3, 5). In TEM micrographs of such areas rupturing of the cell membrane may occasionaly be observed (Fig. 4). Though the plump protrusions are rather scarce, they appear in all filaments studied. They might indicate some secretory or even some pathological process, but a preparation artefact cannot be entirely excluded. Combining light and transmission electron microscopy a few efferently located AVA's were traced from the efferent filament artery to the central venous sinus of the gill filaments. It should be mentioned here that it proved to be unexpectedly difficult to document the whole course of an individual AVA in rainbow trout gill filaments. The reason was the fact that in this species the AVA's apparently never take a straight course and never seek to reach the nearest possible part of the adjacent CVS, as they usually do in T. mossambica (Vogel et al., 1973). All AVA's whose course we tried to follow from the EFA to the CVS run partly parallel to the EFA, partly perpendicular to it for a fairly long distance (4(L60 txm) and often very close to the CVS. They opened preferably into that part of the sinus which was adjacent to the filament cartilage rather than into one located in close vicinity of the EFA (cf. Fig. 1).
Fig. 6. Cross section of AVA. The AVA lumen is extremely narrowed by the endothelial cells (se) and their microvilli. E F A lumen of efferent filament artery, ae arterial endothelial cell with numerous Weibel-Palade-bodies (WP), s m smooth muscle cells of the arterial wall, cc cover cells of the AVA, g c pale interstitial cells with electron dense granules (9,000: 1)
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
379
380
W. Vogel et al.
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
381
Fig. 8. AVA immediately before opening into the CVS (cf. also Fig. 9). Only a short section of the AVA lumen (/) is visible. The two AVA-specific endothelial cells (se) which reach the CVS lumen are still connected by maculae adherentes. Microvilli are scarce. AVA endothelia are in immediate contact with the flat CVS endothelial cells (re). Note the disappearance of a continuous cover cell sheath around the AVA. cc cover cell, ic interstitial cell, iep interstitial endothelial foot processes (9,000:1)
Fig. 7. Luminal labyrinth of AVA. Microvilli of AVA-specific endothelial cells (se) with thin filaments (/") reaching deep down into the cytoplasm. There are plenty of small granules (most probably glycogen) in the AVA endothelia. Both cover cells (cc) and endothelial cells (se) are particularly rich in micropinocytotic vesicles (3,000: 1)
382
W. Vogel et al.
Fig. 9. The luminal area of the AVA (/) is already in contact with that of the CVS (cf. Fig. 8). The two AVA-specific endothelial cells (se) bulging into the CVS still adhere at one point. Note the typically narrow diameter of the AVA (30,000: 1)
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
383
Fig. 10. Afferent side of gill filament sectioned longitudinally. Afferent lamellar arterioles (a/a) originate in fairly regular intervals from the lumen of the afferent filament artery (AFA). No AVA-like structure can be discerned in this artery, rl respiratory lamellae. (360: 1)
In semithin sections, stained with toluidine blue, AVA's cut longitudinally, look like a row of dark cells, while in cross sections only a tight cell clot is visible. Their lumen cannot be clearly recognized in the light microscope though the trout gills examined have been fixed by vascular perfusion. Thus semithin sections could never suffice in order to reveal the true nature of the AVA's. With the exception of the area immediately before the opening into the CVS, each cross sectioned AVA can be clearly discerned from neighbouring nutritive vessels in the TEM by the characteristic bulky endothelial cells with numerous microvilli at their luminal surface (Fig. 6). The microvilli contain straight bundles of filaments which reach fairly deep down into the cytoplasm (Fig. 7). Filament whorls, however, such as in Type I AVA-cells of T. mossambica (Vogel et al., 1974) could not be detected in any AVA cell of S. gairdneri. Thus no clear cut morphological difference exists between cells in the arterial and those in the venous part of an AVA in this species. All AVA endothelia contain a fair though quite variable amount of glycogen granules (Fig. 7). The lumen of the AVA's proved to be always narrowed to a small slit or even to a labyrinth by the endothelial cells and their processes (Figs. 6-8).
384
W. Vogel et al.
In 1 out of 14 AVA's investigated by TEM a red blood cell was apparently trapped, thus indicating some blood flow in the AVA's under certain conditions. Outside the vascular basal lamina of the AVA's a single layer of cover cells is characteristic (Figs. 4, 6). Interstitial endothelial and cover cell processes may also be seen occasionally (Fig. 9). Nerve fibres with mostly unmyelinated axons, can be found in close vicinity in some sections of any AVA when it is traced along its course. In addition comparatively big cells with a various number of electron dense granules scattered throughout the pale cytoplasm are regularly seen in the neighbourhood of the AVA's in S. gairdneri (Figs. 4, 6). The opening of the AVA into the CVS is characterized again by protruding AVA specific endothelial cells (Figs. 8, 9) while CVS endothelial cells are small and insignificant. The ensheathing cover cells of the AVA's usually do not reach quite up to the CVS.
Discussion Arterio-venous anastomoses, connecting the efferent filament artery with the central venous sinus in gill filaments of S. gairdneri have been documented by this study. As in T. mossambica (Vogel et al., 1974), AVA-specific endothelial cells are well discernable from other endothelial cells by their microvillous surface. Apart from ultrastructural details, the only essential difference between the efferently located AVA's of T. mossambica and those of S. gairdneri seems to lay in their length. The fact that the 14 AVA's examined in S. gairdneri never established a fairly short connection between the EFA and CVS, but run very close to the CVS for unexpectedly long distances, made it a rather frustrating task to follow the whole course of an AVA from the EFA to the CVS. It appeared however essential, to prove this course, as any opening into the CVS could also come from nutritive vessels or could be a lock-like connection between separate compartments of the CVS as in T. mossambica. In serial semithin sections, stained with toluidine blue, the course of an AVA can be well followed with the light microscope, but the actual opening into the CVS was never identified unambiguously. This was only possible by TEM examination of such an area. A very careful combination of semithin and ultrathin sectioning was therefore necessary to answer the question as to the kind of connection between AVA's and the CVS. If a concept of AVA functions in S. gairdneri is to be developed one will have to take the length of these vessels and their course as well as their small luminal diameter and the numerous endothelial microvilli into account. We were not able to localize any AVA in the afferent filament region, but it should be emphasized that this does not absolutely rule out their existence. This question cannot be solved by the techniques applied in this study as they are not appropriate for thoroughly screening the afferent lamellar arterioles, where most afferently located AVA's, if such are present at all, may originate. The only method to obtain reliable results in this respect will be serial sectioning of a number of gill filaments over a substantial distance. Injection cast techniques
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
385
could also be considered for this purpose, but due to the extremely narrow lumen of the AVA's it would be difficult to exclude technical reasons when negative results are obtained. Nevertheless it appears reasonable, on the basis of the present results, to expect that in the rainbow trout, most (if not all) AVA's are located at the efferent side of the gill filaments. As in T. mossambica, this would rule out the possibility of substantial intrafilamental afferent-efferent blood shunting in S. gairdneri gills. In other fish species, however, the distribution of afferently and efferently located AVA's in gill filaments may be quite different. It is therefore well possible that various patterns of gill blood flow can be realized in different teleosts.
References Cameron, J.N.: Evidence for the lack of by-pass shunting in teleost gills. J. Fish. Res. Board Canad. 31,211-213 (1974) Cameron, J.N., Polhemus, J.A.: Theory of CO2 exchange in trout gills. J. exp. Biol. 60, 183-194 (1974) Fromme, H.G., Pfautsch, M., Pfefferkorn, G., Bystricky, V.: Die ,,Kritische Punkt" - Trocknung als Pr~iparationsmethode ffir die Raster-Elektronenmikroskopie. Microscopica Acta 73, 29-37 (1972) Hughes, G.M., Morgan, M.: The structure of fish gills in relation to their respiratory function. Biol. Rev. 48, 419-475 (1973) Morgan, M. : Development of secondary lamellae of the gills of the trout, Salmo gairdneri (Richardson). Cell Tiss. Res. 151, 509-523 (1974) Morgan, M., Tovell, P.W.A. : The structure of the gill of the trout, Salmo gairdneri (Richardson) Z.Zellforsch. 142, 147-162 (1973) Newstead, J.D.: Fine structure of the respiratory lamellae of teleostean gills. Z. Zellforsch. 79, 396-428 (1967) Richards, B.D., Fromm, P.O. : Patterns of blood flow through filaments and lamellae of isolated perfused rainbow trout (Salmo gairdneri) gills. Comp. Biochem. Physiol. 29, 1063-1070 (1969) Steen, J.B., Kruysse, A.: The respiratory function of teleostean gills. Comp. Biochem. Physiol. 12, 127-142 (1964) Venable, J.H., Coggeshall, R.: A simplified lead citrate stain for use in electron microscopy. J. Cell Biol. 25, 407-408 (1965) Vogel, W., Vogel, V., Kremers, H.: New aspects of the intra-filamental vascular system in gills of a euryhaline teleost, Tilapia mossambica. Z. Zellforsch. 144, 573-583 (1973) Vogel, W., Vogel, V., Schlote, W.: Ultrastructural study of arterio-venous anastomoses in gill filaments of Tilapia mossambica. Cell Tiss. Res. 155, 491-512 (1974) Watson, M.L.: Staining of tissue sections for electron microscopy with heavy metals. J. biophys. biochem. Cytol. 4, 475-478 (1958)
Received October 2, 1975
Note Added in Proof: Similar results, concerning the existence and the course of AVA's in trout gill filaments have obviously been obtained by Gannon etal. (31st Ann. Proc. Electron Microscopy Soc. Amer. 31,442-443, 1973) with an injection cast technique.
Cell and Tissue Research 9 by Springer-Verlag 1976
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments A Scanning and Transmission Electron Microscopic Study* W. Vogel, V. Vogel, and M. Pfautsch Institutes of Anatomy and of Pathology, University of Tiibingen and Institute for Medical Physics, University of MOnster, Germany
Summary. The origin of arterio-venous anastomoses, connecting the efferent filament artery (EFA) with the central venous sinus (CVS) o f gill filaments can be well discerned by scanning electron microscopy in the rainbow trout. Corresponding vessels between the afferent filament artery and the CVS could not be detected with the techniques applied. AVA-specific endothelial cells are characterized by their bulky shape and their microvillous surface. The general m o r p h o l o g y o f A V A ' s in Salmo gairdneri is very similar to that of A V A ' s in Tilapia mossambica (Vogel et al., 1974) but they are much longer in the trout. N o filament whorls have been encountered in A V A endothelia of Salmo gairdneri. Key words: Gills - Salmo gairdneri - Specialized endothelia venous anastomoses - Electron microscopy.
Arterio-
Introduction The general arrangement and ultrastructure of arterio-venous anastomoses (AVA's) in gill filaments o f a cichlid fish, Tilapia mossambica, has been described recently (Vogel et al., 1973, 1974) 1. One characteristic feature of these A V A ' s in T. mossambica are highly specialized endothelial cells with numerous tentacular microvilli at their surface. These AVAospecific cells usually protrude into the lumen o f the corresponding gill filament artery at the anastomosis' origin. It is therefore possible to demonstrate the origin of A V A ' s in longitudinally sliced efferent gill filament arteries in T. mossambica by scanning electron microscopy (unpublished observations).
Send offprint requests to: Dr. Walter O.P. Vogel, Anatomisches Institut der Universitfit Tiibingen, D-7400 T/ibingen, Osterbergstr. 3, Federal Republic of Germany * This study is dedicated to Prof. Dr. W. Graumann, Director of the Institute of Anatomy, University of Tiibingen, on the occasion of his 60th birthday. It was supported by the Deutsche Forschungsgemeinschaft 1 In this species most of the AVA's are located between the efferent filament artery and a central venous sinus (CVS) while those between the afferent filament artery and the CVS are extremely rare
374
W. Vogel et al.
~E
b|e--~L~ /
fe
It
pc
CVS
\|174 e~'
a[a .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
~A~
Fig.
1. Diagram of cross sectioned gill filament. Sections for T E M were cut in this plane. The plane of sectioning for the SEM preparations was perpendicular to it where indicated by broken lines at the afferent (A) and at the efferent (E) side of the filament. Thick arrows: directions of scanning. Thin arrows: Main blood flow direction in the respiratory lamellae. 9 usual site of AVA origin in the wall of the efferent filament artery (EFA). ** region of the central venous sinus (CVS) where all of the A V A ' s investigated in this respect opened into the CVS. AFA afferent filament artery, ala afferent lamellar arteriole, ble basal lamellar endothelial cells, ela efferent lamellar arteriole, fc filament cartilage, fe filament epithelium, ll lamellar lacuna (lamellar blood space), pe pillar cells
In spite of the pioneer work of Steen and Kruysse (1964) in this field it is still controversial, whether similar vascular connections do exist in gill filaments of other teleosts. Investigations carried out mainly on salmonids (Steen and Kruysse, 1964; Newstead, 1967; Richards and Fromm, 1969; Morgan and Tovell, 1973; Morgan, 1974; Cameron, 1974; cf. also Hughes and Morgan, 1973; Cameron and Polhemus, 1974) have yielded contradictory results. As such AVA's in fish gill filaments may be important for respiratory and for
Fig. 2. Efferent side of gill filament sectioned longitudinally: Luminal view of efferent filament artery (EFA) and of three efferent lamellar arterioles (ela). Two of the latter share a c o m m o n outflow to the EFA. The origin of two arterio-venous anastomoses is well recognizable by the A V A specific endothelial cells (se) bulging between the endothelial cells of the artery. A few of the recessus of the central venous sinus (CVS) reaching far around the arterial vessels still contain some red blood cells (rbc). Arrows indicate blood flow direction (630: 1)
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
375
Fig. 3. Closer view of an AVA-origin in the wall of the efferent filament artery, a e arterial endothelial cell, s e AVA-specific endothelia with bulky cell projections (arrows), l lumen of the AVA (2,750: 1)
376
W. Vogel et al.
osmoregulatory gill functions, we decided to search again for these anastomoses in gill filaments of Salmo gairdneri, applying mainly scanning electronmicroscopic techniques. Material and Methods The fish, Salmo gairdneri (Richardson), were purchased from a local dealer and kept in well oxygenated running tap water (14~ for up to 14 h prior to use. 3 fish, 25-35 cm long, weighing 310450 g, were studied. Each one was anaesthetized with MS 222 (0.5 g/l) for 2 min. The vascular system of the fish was perfused through the ventral aorta with the following perfusion mixtures: Ringer solution (pH 7.3; osmolality: 320 mOsm/Kg H20 ) for 0.5 min, followed immediately by a 2% glutaraldehyde solution buffered with Na-cacodylate (pH 7.3; osmolality: 430 mOsm/Kg H20). The perfusion pressure was adjusted to 40q50 Torr. After 10 min of perfusion, gill filaments from the middle of the first gill arches of both sides were excised and immersed in a fresh portion of the same fixative for 3-4 h at 4~ The tissue was then rinsed in cold cacodylate buffer supplemented with saccharose (osmolality: 320 mOsm/Kg H20), postfixed in 2% OsO4 adjusted with Na-cacodylate buffer to pH 7.3, for 2 h and partially dehydrated in a graded ethanol series in steps of 10% starting with 30% ethanol. In 70% ethanol the material was divided into 2 groups:
1. Scanning Electron Microscopy ( S E M )
Immersed in a fresh cold ethanol solution (70%) the efferent or afferent side of single gill filaments was cut longitudinally by means of a razor blade. Under the control of a dissecting microscope this knife was adjusted with a micromanipulator such, that sectioning took place through the middle of either the afferent or efferent filament artery (cf. Fig. 1). Thus fairly long pieces of these vessels could be opened. The specimens were then processed further for drying above the critical point of Freon 13 (Fromme et al., 1972). Dried specimens were mounted on aluminium specimen studs with conducting silver paint and coated with gold by means of a Leitz sputter coating unit. They were examined in a scanning electron microscope "Stereoscan MK I " (Cambridge Instruments Ltd.).
2. Transm&sion Electron Microscopy ( T E M )
Whole filaments were further dehydrated through graded ethanol to propylene oxide and embedded in Araldite. Semi- and ultrathin sections were cut with a Reichert OmU3 ultramicrotome. The plane of sectioning is shown in Figure 1. Semithin sections were stained with toluidine blue for light microscopy. Ultrathin sections stained with uranyl-acetate (Watson, 1958) and lead citrate (Venable and Goggeshall, 1965) were examined with a Siemens Elmiskop 102.
Fig. 4. Oblique section of an AVA at its origin in the efferent filament artery (EFA). AVA-specific endothelia (se) bear microvilli as well as irregular, bulky protrusions occasionally with ruptures of the cell membrane (arrow). ae arterial endothelial cells, sm smooth muscle cells, gc bright cell with electron dense granules of various size, cc cover cells of the AVA in immediate contact with smooth muscle cells of the EFA. ve venous endothelial cell. Parts of the central venous sinus (CFS) are very often in such a close vicinity to the EFA. No AVA, however, has been observed to open immediately into the CVS in this area (6,000: 1) Fig. 5. Origin of an AVA as seen from the lumen of the efferent filament artery, ae arterial endothelial cell, se AVA-specific endothelial cell, arrows: microvilli at the border of normal endothelial cells, p bulky protrusion on the surface of AVA endothelial cell (6,870 : 1)
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
377
378
W. Vogel et al.
Results
The present investigation confirms that arterio-venous anastomoses (AVA's) do exist between the efferent filament artery (EFA) and a sinusoidal central venous system (CVS) of S. gairdneri gill filaments. Most of these AVA's are located in that part of the EFA adjacent to the filament core, where they are easily accessible for SEM examination (cf. Fig. 1). In filamental cross sections, some AVA's could also be seen to originate from efferent lamellar arterioles. Judging from the SEM preparations the AVA's appear at irregular intervals along the wall of the EFA. They seem to be less numerous than the lamellar pairs in the filaments. The microvilli at the surface of AVA-specific endothelial cells, which protrude into the lumen of efferent filament arteries, greatly facilitate the location of AVA's (i.e. their origin) by scanning electron microscopy (Fig. 2). In afferent filament arteries, however, no similar cells of corresponding AVA's have been detected with this technique (Fig. 10). In a few filaments afferent lamellar arterioles have also been investigated, but again no AVA's could be identified here. In the EFA, microvilli are not entirely confined to AVA endothelial cells (cf. Fig. 5), but here they are definitely concentrated and usually longer than in normal endothelia. These microvilli are apparently not evenly distributed over the luminal surface of the AVA cells, but are preferably located around the AVA lumen and also in the area of contact with normal endothelia. In addition to the microvilli, a few, rather plump protrusions can be seen in SEM pictures of AVA endothelia (Figs. 3, 5). In TEM micrographs of such areas rupturing of the cell membrane may occasionaly be observed (Fig. 4). Though the plump protrusions are rather scarce, they appear in all filaments studied. They might indicate some secretory or even some pathological process, but a preparation artefact cannot be entirely excluded. Combining light and transmission electron microscopy a few efferently located AVA's were traced from the efferent filament artery to the central venous sinus of the gill filaments. It should be mentioned here that it proved to be unexpectedly difficult to document the whole course of an individual AVA in rainbow trout gill filaments. The reason was the fact that in this species the AVA's apparently never take a straight course and never seek to reach the nearest possible part of the adjacent CVS, as they usually do in T. mossambica (Vogel et al., 1973). All AVA's whose course we tried to follow from the EFA to the CVS run partly parallel to the EFA, partly perpendicular to it for a fairly long distance (4(L60 txm) and often very close to the CVS. They opened preferably into that part of the sinus which was adjacent to the filament cartilage rather than into one located in close vicinity of the EFA (cf. Fig. 1).
Fig. 6. Cross section of AVA. The AVA lumen is extremely narrowed by the endothelial cells (se) and their microvilli. E F A lumen of efferent filament artery, ae arterial endothelial cell with numerous Weibel-Palade-bodies (WP), s m smooth muscle cells of the arterial wall, cc cover cells of the AVA, g c pale interstitial cells with electron dense granules (9,000: 1)
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
379
380
W. Vogel et al.
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
381
Fig. 8. AVA immediately before opening into the CVS (cf. also Fig. 9). Only a short section of the AVA lumen (/) is visible. The two AVA-specific endothelial cells (se) which reach the CVS lumen are still connected by maculae adherentes. Microvilli are scarce. AVA endothelia are in immediate contact with the flat CVS endothelial cells (re). Note the disappearance of a continuous cover cell sheath around the AVA. cc cover cell, ic interstitial cell, iep interstitial endothelial foot processes (9,000:1)
Fig. 7. Luminal labyrinth of AVA. Microvilli of AVA-specific endothelial cells (se) with thin filaments (/") reaching deep down into the cytoplasm. There are plenty of small granules (most probably glycogen) in the AVA endothelia. Both cover cells (cc) and endothelial cells (se) are particularly rich in micropinocytotic vesicles (3,000: 1)
382
W. Vogel et al.
Fig. 9. The luminal area of the AVA (/) is already in contact with that of the CVS (cf. Fig. 8). The two AVA-specific endothelial cells (se) bulging into the CVS still adhere at one point. Note the typically narrow diameter of the AVA (30,000: 1)
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
383
Fig. 10. Afferent side of gill filament sectioned longitudinally. Afferent lamellar arterioles (a/a) originate in fairly regular intervals from the lumen of the afferent filament artery (AFA). No AVA-like structure can be discerned in this artery, rl respiratory lamellae. (360: 1)
In semithin sections, stained with toluidine blue, AVA's cut longitudinally, look like a row of dark cells, while in cross sections only a tight cell clot is visible. Their lumen cannot be clearly recognized in the light microscope though the trout gills examined have been fixed by vascular perfusion. Thus semithin sections could never suffice in order to reveal the true nature of the AVA's. With the exception of the area immediately before the opening into the CVS, each cross sectioned AVA can be clearly discerned from neighbouring nutritive vessels in the TEM by the characteristic bulky endothelial cells with numerous microvilli at their luminal surface (Fig. 6). The microvilli contain straight bundles of filaments which reach fairly deep down into the cytoplasm (Fig. 7). Filament whorls, however, such as in Type I AVA-cells of T. mossambica (Vogel et al., 1974) could not be detected in any AVA cell of S. gairdneri. Thus no clear cut morphological difference exists between cells in the arterial and those in the venous part of an AVA in this species. All AVA endothelia contain a fair though quite variable amount of glycogen granules (Fig. 7). The lumen of the AVA's proved to be always narrowed to a small slit or even to a labyrinth by the endothelial cells and their processes (Figs. 6-8).
384
W. Vogel et al.
In 1 out of 14 AVA's investigated by TEM a red blood cell was apparently trapped, thus indicating some blood flow in the AVA's under certain conditions. Outside the vascular basal lamina of the AVA's a single layer of cover cells is characteristic (Figs. 4, 6). Interstitial endothelial and cover cell processes may also be seen occasionally (Fig. 9). Nerve fibres with mostly unmyelinated axons, can be found in close vicinity in some sections of any AVA when it is traced along its course. In addition comparatively big cells with a various number of electron dense granules scattered throughout the pale cytoplasm are regularly seen in the neighbourhood of the AVA's in S. gairdneri (Figs. 4, 6). The opening of the AVA into the CVS is characterized again by protruding AVA specific endothelial cells (Figs. 8, 9) while CVS endothelial cells are small and insignificant. The ensheathing cover cells of the AVA's usually do not reach quite up to the CVS.
Discussion Arterio-venous anastomoses, connecting the efferent filament artery with the central venous sinus in gill filaments of S. gairdneri have been documented by this study. As in T. mossambica (Vogel et al., 1974), AVA-specific endothelial cells are well discernable from other endothelial cells by their microvillous surface. Apart from ultrastructural details, the only essential difference between the efferently located AVA's of T. mossambica and those of S. gairdneri seems to lay in their length. The fact that the 14 AVA's examined in S. gairdneri never established a fairly short connection between the EFA and CVS, but run very close to the CVS for unexpectedly long distances, made it a rather frustrating task to follow the whole course of an AVA from the EFA to the CVS. It appeared however essential, to prove this course, as any opening into the CVS could also come from nutritive vessels or could be a lock-like connection between separate compartments of the CVS as in T. mossambica. In serial semithin sections, stained with toluidine blue, the course of an AVA can be well followed with the light microscope, but the actual opening into the CVS was never identified unambiguously. This was only possible by TEM examination of such an area. A very careful combination of semithin and ultrathin sectioning was therefore necessary to answer the question as to the kind of connection between AVA's and the CVS. If a concept of AVA functions in S. gairdneri is to be developed one will have to take the length of these vessels and their course as well as their small luminal diameter and the numerous endothelial microvilli into account. We were not able to localize any AVA in the afferent filament region, but it should be emphasized that this does not absolutely rule out their existence. This question cannot be solved by the techniques applied in this study as they are not appropriate for thoroughly screening the afferent lamellar arterioles, where most afferently located AVA's, if such are present at all, may originate. The only method to obtain reliable results in this respect will be serial sectioning of a number of gill filaments over a substantial distance. Injection cast techniques
Arterio-venous Anastomoses in Rainbow Trout Gill Filaments
385
could also be considered for this purpose, but due to the extremely narrow lumen of the AVA's it would be difficult to exclude technical reasons when negative results are obtained. Nevertheless it appears reasonable, on the basis of the present results, to expect that in the rainbow trout, most (if not all) AVA's are located at the efferent side of the gill filaments. As in T. mossambica, this would rule out the possibility of substantial intrafilamental afferent-efferent blood shunting in S. gairdneri gills. In other fish species, however, the distribution of afferently and efferently located AVA's in gill filaments may be quite different. It is therefore well possible that various patterns of gill blood flow can be realized in different teleosts.
References Cameron, J.N.: Evidence for the lack of by-pass shunting in teleost gills. J. Fish. Res. Board Canad. 31,211-213 (1974) Cameron, J.N., Polhemus, J.A.: Theory of CO2 exchange in trout gills. J. exp. Biol. 60, 183-194 (1974) Fromme, H.G., Pfautsch, M., Pfefferkorn, G., Bystricky, V.: Die ,,Kritische Punkt" - Trocknung als Pr~iparationsmethode ffir die Raster-Elektronenmikroskopie. Microscopica Acta 73, 29-37 (1972) Hughes, G.M., Morgan, M.: The structure of fish gills in relation to their respiratory function. Biol. Rev. 48, 419-475 (1973) Morgan, M. : Development of secondary lamellae of the gills of the trout, Salmo gairdneri (Richardson). Cell Tiss. Res. 151, 509-523 (1974) Morgan, M., Tovell, P.W.A. : The structure of the gill of the trout, Salmo gairdneri (Richardson) Z.Zellforsch. 142, 147-162 (1973) Newstead, J.D.: Fine structure of the respiratory lamellae of teleostean gills. Z. Zellforsch. 79, 396-428 (1967) Richards, B.D., Fromm, P.O. : Patterns of blood flow through filaments and lamellae of isolated perfused rainbow trout (Salmo gairdneri) gills. Comp. Biochem. Physiol. 29, 1063-1070 (1969) Steen, J.B., Kruysse, A.: The respiratory function of teleostean gills. Comp. Biochem. Physiol. 12, 127-142 (1964) Venable, J.H., Coggeshall, R.: A simplified lead citrate stain for use in electron microscopy. J. Cell Biol. 25, 407-408 (1965) Vogel, W., Vogel, V., Kremers, H.: New aspects of the intra-filamental vascular system in gills of a euryhaline teleost, Tilapia mossambica. Z. Zellforsch. 144, 573-583 (1973) Vogel, W., Vogel, V., Schlote, W.: Ultrastructural study of arterio-venous anastomoses in gill filaments of Tilapia mossambica. Cell Tiss. Res. 155, 491-512 (1974) Watson, M.L.: Staining of tissue sections for electron microscopy with heavy metals. J. biophys. biochem. Cytol. 4, 475-478 (1958)
Received October 2, 1975
Note Added in Proof: Similar results, concerning the existence and the course of AVA's in trout gill filaments have obviously been obtained by Gannon etal. (31st Ann. Proc. Electron Microscopy Soc. Amer. 31,442-443, 1973) with an injection cast technique.
