Microvascular anatomy of the brain of the adult pipid frog, Xenopus laevis (Daudin): A scanning electron microscopic study of vascular corrosion casts

Abstract To demonstrate the 3D microvascular anatomy of the brain of the model organism Xenopus laevis Daudin scanning electron microscopy of vascular corrosion casts was correlated with light microscopy of stained 7 µm thick serial tissues sections. Results showed that supplying arteries descended from the leptomeningeal surface without remarkable branchings straight to the subventricular zone where they branched and capillarized. Capillaries showed few H‐ and/or Y‐shaped anastomoses during their centrifugal course toward the leptomeningeal surface where they drained into cerebral venules and veins. Apart from the accessory olfactory bulb and the vestibule‐cochlear nucleus where capillaries were densely packed, capillaries formed a wide‐meshed 3D network throughout the brain parenchyma and thus contrasted to urodelian brains where hairpin‐shaped capillaries descend from the leptomeningeal vessels into varying depths of the brain parenchyma. In about two‐third of specimens, a closed arterial circle of Willis was found at the base of the brain. If this circle in Xenopus might serve the same two functions as in men is briefly discussed. Choroid plexuses of third and fourth ventricle were found to have a high venous, but a low arterial inflow via one small choroidal artery only. Findings are compared with previous studies on the vascularization of the anuran brain and discrepancies in respect to presence or absence of particular arteries and/or veins in Ranids, Bufonids, and Pipids studied so far are discussed with particular emphasis on the techniques used in the various studies published so far.


| I N TR ODU C TI ON
The vertebrate brain cannot store energy and thus relies on a permanent supply with nutrients and oxygen via the circulatory system. Presently, we have a detailed knowledge of anatomy and efferent and afferent neural connections of the functional systems of the anuran brain (for reviews see e.g., Kemali & Braitenberg, 1969;Kuhlenbeck, 1977;Llinas & Precht, 1976;Ten Donkelaar, 1998). Comparatively little, however, is known on its detailed microvascular anatomy. Studies mainly focus on gross arterial supply and venous drainage (Abbie, 1934;Craigie, 1938;Gaupp, 1899;Gillilan, 1967;Millard, 1940Millard, , 1945Millard, , 1949Sch€ obl, 1882;Socha, 1930) or describe the vascularization of specific brain areas only (Cruz, 1959;Dierickx, Goossens, & De Waele, 1970Dierickx, Lombaerts-Vandenberghe, & Druyts, 1971;Goossens, Dierickx, & De Waele, 1973;Rodriguez & Pizzi, 1967). In most studies, authors use India-ink or India ink-gelatin injections with subsequent clearing of the brain tissue according to Spalteholz (1914) and study these specimens either in toto using the stereomicroscope or analyze serially thick-sectioned brains by conventional light microscopy (LM). In India-ink injected specimens, it is difficult to positively differentiate arteries from veins and vessels have to be followed from their identifiable origin. Though differentiation of vessel nature can be improved by the double injection technique (Ambach & Palkovits, 1974), the low depth of focus of the light microscope requires amendment by serial sectioning and laborious reconstruction work if information about the entire cerebrovascular system is needed. Staining of vessels by perfusion with lipophilic DiI (vessel painting; Hughes, Dashkin, & Defazio, 2014) followed by observation with conventional fluorescence microscopy or confocal microscopy improves spatial resolution and depth of focus, but due to the limited depth of penetration of the laser this method still relies on thick sectioning of larger brains.
The availability of resins which enable to cast the entire vascular bed from the aortic trunk(s) through the capillaries to the opened heart (Taniguchi, Ohta, & Tajiri, 1952) and the application of the scanning electron microscope (SEM) to study these resin-made vascular casts (Murakami, 1971) enable to document the 3D arrangement of blood vessels with a high depth of focus and a high spatial resolution. The possibility to differentiate arterial and venous vessels by means of their characteristic endothelial cell nuclear imprints on cast surfaces (Miodonski, Hodde, & Bakker, 1976) and the application of 3D morphometry (Malkusch, Konerding, Klapthor, & Bruch, 1995;Minnich, Leeb, Bernroider, & Lametschwandtner, 1999) even allow to gain quantitative data on physiologically relevant vascular parameters like vessel diameters, lengths, and branching angles (St€ ottinger, Klein, Minnich, & Lametschwandtner, 2006).
To date, we have a 3D, high-resolution atlas of normal vascular development in the embryo (Levine, Munoz-Sanjuan, Bell, North, & Brivanlou, 2003) and some knowledge of the angiogenesis within the optic tectum of the embryo, tadpole, and postmetamorphic Xenopus laevis (Rovainen & Kakarala, 1989;Tiedeken & Rovainen, 1991). However, we still lack an in-depth knowledge of the brain's 3D microvascular anatomy of this model organism in biological research. A profound knowledge of microvascular patterns and in particular of vascular connections is crucial to better understand energy supply and distribution within neuroanatomically clearly defined brain areas. Here, we demonstrate that SEM of vascular corrosion casts in combination with LM of stained serial tissue sections enables to visualize minute details of the microvascular bed and to topologically attribute them to distinct small brain areas. Additionally, we show that color-coding of arteries (red), veins (blue), and meningeal vessels (green) in SEM micrographs distinctly facilitates identification of vessel origins, courses, branching patterns, and areas of supply and drainage.

| Animals
Ten adult animals (four males, 34-43 g, body length: 7.5-8.0 cm; six females; 33-100 g, body length: 6.3-10.0 cm) of the pipid frog, X. laevis (Daudin) were studied. The larger animals were purchased from Horst Kaehler (Hamburg, Germany), the smaller ones were raised in our animal facility. Animals were housed in aquaria (tap water; depth: 15 cm) equipped with aquarium filters and fed twice a week with either dried Gammarus pulex or grinded beef heart.

| Histomorphology
One male (body weight: 77.0 g, body length: 8.0 cm) and one female animal (body weight: 78 g, body length: 10.0 cm) were killed by immersion into an aqueous solution of MS 222 (0.5%; Sigma-Aldrich Chemie, Steinbuch, Germany). After, weighing animals were pinned in supine position on a wax plate. The heart with bulbus cordis and truncus arteriosus was exposed by thoracotomy and a ligature was placed around the bulbus cordis. Next, the ventricle was cut open and a blunt grinded vein flow G19 (Braun, Melsungen, Germany) guided by a micromanipulator was inserted through the opened ventricle into the truncus arteriosus. Subsequently, the blunted needle was tied in place with a ligature from thread to ensure its stability during the following rinsing and fixing processes. Finally, the sinus venosus was cut open to allow efflux of blood and rinsing with amphibian ringer solution (Adam & Czihak, 1964) started. The flow rate of the infusion pump (Habel, Vienna) was set to 40 mL/hr. When clear reflux drained from the opened sinus venosus fixation with 10 mL Bouin's solution (Adam & Czihak, 1964) was started using the same flow rate. The fixed brain was removed from the brain cavity, dehydrated, and embedded in paraplast. One series each of 7 mm thick transverse and of longitudinal sections were LAMETSCHWANDTNER AND MINNICH | 951 stained according to Goldner (Adam & Czihak, 1964). Tissue sections were analyzed with an Olympus X51 microscope. Images were recorded by a Color View III digital camera (Soft Imaging Systems, FRG). If necessary brightness and contrast of images were adjusted using Photoshop 7.0 (Adobe Inc., Redwood, CA).

| Vascular corrosion casting
Seventeen adult X. laevis (4 males, 13 females; body weights: 39.0-93.0 g, total lengths: 75-95 mm) were studied. For euthanasia and rinsing, see Section 2.2. When clear reflux drained from the opened sinus venosus 10 ml of Mercox CL-2B (Dainippon Ink and Chemicals, Tokyo, Japan; Ladd Burlington, Vermont, USA) diluted with monomeric methyl methacrylate (4 1 1, v 1 v, 10 mL monomeric methyl acrylate contained 0.85 g initiator paste MA) were injected with the infusor (see Section 2.2) at a flow rate of 41 mL/hr. When the effluent resin became viscous (after 13-14 min) or the whole amount of resin had been perfused the injection was stopped and the animals were left for about 30 min at room temperature to allow hardening of the injected resin. Animals then were put into a water-bath (608C; 12-24 hr) to temper the injected resin. Next, specimens were macerated in potassium hydroxide (7.5%; 408C; 2-24 hr), rinsed three times in distilled water, submerged in 2% hydrochloric acid, rinsed three times in distilled water followed by submersion in formic acid (5%; 208C; 5-15 min) to dissolve any residual organic matter adhering to the cast surfaces. Finally, specimens were rinsed another three times in distilled water and frozen in fresh distilled water. Ice-embedded casts were freeze-dried in a Lyovac GT2 (Leybold-Heraeus, Cologne, Germany). Brains were excised and mounted onto specimen stubs using the "conductive bridge-method" (Lametschwandtner, Miodonski, & Simonsberger, 1980), either evaporated with carbon and gold and/or sputter-coated with gold, and examined in the SEM ESEM XL-30 (FEI, Eindhoven, The Netherlands) at an accelerating voltage of 10 kV.
Individual cast brains were either mounted in toto onto specimen stubs for detailed SEM analyses or were cut transversely or sagittally.
For this purpose, brain casts were submerged into distilled water, frozen and sectioned while embedded in ice using a mini-wheel saw placed in the cryo-chamber of a cryo-microtome . Sectioned casts where cleaned, frozen, freeze-dried, mounted, and analyzed in the SEM.
In some specimen's course, branching patterns and areas of supply (or drainage) of individual vessels were exposed by ripping-off overlaying vessels under binocular control by fine tipped insect pins.

| R E SU LTS
Stereomicroscopic inspection of whole brain vascular corrosion casts revealed excellent filling of pial vessels. Subsequent SEM of sectioned or partially dissected casts confirmed that also intraparenchymal microvessels were fully replicated and imposed as a 3D network whose density was highest in the accessory olfactory bulb and the vestibulocochlear nucleus, but low in the remaining brain areas.
To unravel the 3D vascular network of the entire brain and spinal cord, we first report gross arterial supply. Starting with extra-and intracranial feeding vessels, we follow their branches intraparenchymally toward the microvascular bed. Next, we trace gross venous drainage routes toward the pial surface and finally analyze microvascular patterns.

| Gross arterial supply
The brain of adult X. laevis is bilaterally fed via (a) common carotid artery-internal carotid artery-cerebral carotid artery and (b) the vertebral artery which connects via the communicating artery with the single basilar artery (Figures 1-3 Here, it branches into rostrally, laterally, and caudally directed branches which supply the infundibular lobe in a centrifugal manner (Figure 7).
The preoptic artery arises either from the medial, dorsal, or lateral aspect of the anterior branch of the cerebral carotid artery. It supplies part of the preoptic area and caudal pallial and septal areas of the telencephalon (Figures 1-10, and 11).
A prominent septal artery was found in a few specimens only (Figures 2 and 12214). Generally, this artery ascends without any branching through the striatum toward the subventricular zone where it branches intensively (Figures 12-14). Branches supply striatal, pallial, and septal areas (Figures 12-14).
In most specimens studied, right and left anterior branches of the cerebral carotid arteries join at about the caudal end of the ventral interhemispheric fissure (Figures 1 and 2). At this site or slightly anteriorly or posteriorly medial and lateral olfactory arteries and the posterior telencephalic artery arise. While the former run rostrally to supply olfactory and accessory olfactory bulbs, the latter ascends between the telencephalic hemispheres toward the dorsal interhemispheric fissure ( Figures 15 and 16). The artery either ascends vertically ( Figure 15; see Note the superficial infundibular artery (sia) which gives off a medially directed branch (large arrows) supplying the delicate capillary network of the retrochiasmatic (rostral) and infundibular (caudal) region. The parent artery bends toward caudal to supply the median eminence (me) via a medially directed branch (arrowheads) and the intermediate lobe of the hypophysis (ilhy) via a laterally directed terminal branch (small arrows) FIGURE 5 Same as Figure 4 but after removal of vascular beds of infundibular region (ventral hypothalamus) and hypophysis. Note the deeply penetrating branches of the superficial infundibular arteries (sia; arrowheads), the retroinfundibular communicating artery (rica) and the origin of prominent ascending rostral tegmental arteries (rta). Inset a. Histomorphology of (right) optic tectum (ot) and tegmentum (teg) of the mesencephalon. Paraplast embedded Goldner stained tissue section (7 mm). Transverse section at the level of the ascending rostral tegmental arteries (arrowheads). Arrow marks the mesencephalic ventricle. ir, infundibular recess. Inset b. Same as inset a, but slightly more caudal section. Note the horizontally running caudal branch of the (right) rostral tegmental artery (arrow) and its ascending branch (arrowhead) FIGURE 6 Arterial supply and venous drainage of the dorsal hypothalamus. Note the arterial supply via the deep infundibular artery (dia) and subependymally located rostral branches of rostral tegmental arteries (rta). The diencephalic vein (dv) drains rostral areas while the hypothalamic vein (hv) drains caudal areas For origin, course, branching pattern, and areas of supply of the deep infundibular artery, see above.
The diencephalic artery either arises individually or with a common stem together with the anterior inferior mesencephalic artery In most specimens, the mesencephalon bilaterally owns one inferior and two superior mesencephalic arteries 18,19,21,(24)(25)(26)(27)(28)(29)and 31) and several tegmental arteries (Figures 28 and 29). The anterior inferior mesencephalic artery-which also could be termed a tegmental artery-arises either from the diencephalic artery ( Figure 18 In most specimens, a retroinfundibular communicating artery connects right and left posterior branches of the cerebral carotid arteries (Figures 4-7). The caliber of this artery often varies greatly between its right and left sided portion. In few cases, the middle portion of this artery is very thin or is even absent and right and left stems of the artery do not interconnect at the midline (Figures 6 and 7). In this case, these arteries directly continue as the most rostral (mesencephalic) tegmental arteries (Figures 6 and 7). Tegmental arteries ascend in a rostroor caudo-dorsal direction. In general, they bifurcate into a rostrally and caudally directed branch. The rostral branch supplies caudal areas of dorsal hypothalamus and thalamus ( Figures 28 and 29), the caudal branch supplies the mesencephalic tegmentum and ventral and lateral areas of the optic tectum ( Figure 29).
FIG URE 7 Same specimen as in Figure 6, but after exposure of the deep infundibular artery by removal of overlaying vessels. Note that the artery runs without branching straight toward the subependymal zone where it branches into rostrally and caudally directed branches FIG URE 8 Internal microvascular anatomy of the brain of adult X. laevis. Sagittally sectioned vascular corrosion cast. Right half of the brain displaying telencephalon (tel), diencephalon (di) (without hypothalamus and hypophysis), mesencephalon (mes), cerebellum (cer), and rhombencephalon (rho). Asterisks mark conductive bridges   Obliquely ascending posterior telencephalic artery (pta) giving off a rostrally directed branch at the level of the dorsomedial pallium (arrowhead). The main trunk bends toward caudal and issues several branches (arrows). Inset a. Characteristic endothelial cell nuclei imprints (arrowheads) at the surface of a cast artery (posterior telencephalic artery). Imprints are longish and orientate parallel to the vessel axis. Inset b. Characteristic endothelial cell nuclei imprints (arrowheads) at the surface of a cast vein (interhemispheric vein). Imprints are oval to roundish and orientate randomly  (tel), epithalamus (epi), thalamus (th), optic tectum (ot), mesencephalic, and rhombencephalic tegmentum (teg). Lateral view. Rostral is to the right. Note the small caliber of the diencephalic artery (da) which shares a common stem with a lateral tegmental artery (arrow). The posterior superior mesencephalic artery (psma) shares a common stem with the inferior mesencephalic (tegmental) artery (arrowhead) FIG URE 19 Microvascular anatomy of caudal dorsal (dp) and lateral pallium (lp), epithalamus (epi) and optic tectum (ot). Dorsolateral view. Note the prominent oblique cranial vein (ocv) and the medially located longitudinal mesencephalic vein (lmv) which by two tributaries (arrows) drains epithalamic, thalamic and rostral, and lateral areas of the optic tecta

| Gross venous drainage
Course, calibers, and branching patterns of right and left cerebral drainage routes may greatly vary. In detail, olfactory bulbs, accessory olfactory bulbs, telencephalon, and rostral thalamic areas drain via prominent lateral telencephalic veins and a single interhemispheric vein (Figures 12, 15, and 16).     (Figures 31 and 40). The rhombencephalic tegmentum drains via lateral rhombencephalic veins which take their origin at the ventral midline ( Figure 43). Veins ascend around the lateral tegmental surface and join with neighboring veins to form larger trunks which further ascend toward the margins of the rhombencephalic fossa (Figures 18 and 30).
In the rostral area, they directly drain into the choroid plexus of the fourth ventricle via short ascending veins, in the caudal regions they mostly drain into a vein which runs along the dorsal margin of the rhombencephalic fossa toward rostral (Figures 30 and 41). This vein then also drains into the choroid plexus of the fourth ventricle.
The choroid plexus topping the rhombencephalic fossa varies interindividually in respect to shape, size, and vascular patterns (Figures 8,9,21,41,and 42). In most cases, it has a shape like an isosceles triangle with the tip pointing caudally (Figures 21). In general, the plexus covers the entire (Figure 21), but in rare cases, it covers only part of the rhombencephalic fossa ( Figure 41). The plexus forms from transversely orientated sheet-like vascular lamellae (Figures 44 and 45).
These lamellae consist of interconnected wide sinusoids with a somewhat thicker vessel at the ventral margins ( Figure 45). In the rostral areas, lamellae are slightly inclined toward rostrally in the caudal portion they incline toward caudally (Figures 8 and 9). The tela of the choroid plexus IV shows a dense network of sinusoids of different shapes and sizes (Figures 21, 41, and 42). Several choroidal veins drain the plexus into the oblique occipital veins which run aside the lateral margins of the plexus toward rostro-laterally (Figures 21, 41, and 42). In most cases, the oblique occipital veins form by the bifurcation of the dorsal spinal vein at the caudal tip of the plexus and empty into the   (Figure 21). In a few cases, the dorsal spinal vein bifurcates more caudally and the two branches approach the caudal margin of the choroid plexus more laterally (Figure 41).  Millard (1940) described the vascular anatomy of X. laevis. In her impressive study, she performs careful dissections and follows origin, course, and areas of supply of main cerebral vessels. Due to the limited depth of focus of the stereomicroscope, her study gives no information about the microvascular territories. SEM of vascular corrosion casts (Aharinejad & Lametschwandtner, 1992;Lametschwandtner, Lametschwandtner, & Weiger, 1990;Motta, Murakami, & Fujita, 1992;Murakami, 1971) overcame this limitation and for the first time, the 3D FIG URE 36 Gross arterial supply and venous drainage of olfactory bulb (ob), telencephalon (tel) and diencephalon (di). Ventro-lateral view. Rostral is to the right. Note the lateral telencephalic vein (ltv) and its dorsal and ventral tributaries. Inset a. Confluence of cerebral veins at the level of the prootic foramen. Lateral view FIG URE 37 Course and caliber of the (right) oblique cranial vein (ocv). Lateral view at the optic tectum (ot), diencephalon (di), and caudal telencephalon (tel). Rostral is to the right FIGURE 38 Anatomy of the encephalo-posthypophysial portal system. Ventral aspect. Hypophysis and caudal infundibulum are removed. Note the hypothalamic branch (hb) and the ventral tegmental branches (vtb) of the portal system LAMETSCHWANDTNER AND MINNICH | 963 microvascular anatomy of the anuran (B. bufo; Albrecht et al., 1978;Lametschwandtner & Simonsberger, 1975;Lametschwandtner et al., 1976Lametschwandtner et al., , 1977aLametschwandtner et al., , 1977bLametschwandtner et al., , 1977cLametschwandtner et al., , 1978Lametschwandtner et al., , 1979aLametschwandtner et al., , 1979bLametschwandtner et al., , 1980aLametschwandtner et al., , 1980b, the urodelian brain (T. cristatus and T. carnifex; Lazzari et al., 1991), and A. mexicanum (Lazzari & Franceschini, 2003) was shown. The spatial resolution of the SEM is sufficiently high to clearly differentiate arteries from veins by means of the characteristic endothelial cell nuclei imprint patterns displayed on the surfaces of cast vessels (Miodonski et al., 1976; see also Figure 16, insets a and b) and enables to demonstrate origin, caliber, course, and branching patterns of individual blood vessels throughout the whole organ. If this technique is further supplemented by corresponding tissue sections with subsequent histomorphological analyses areas of supply and drainage of individual blood vessels can be defined with high reliability. Sectioning of vascular corrosion casts and/or removal of vessels layer by layer by using fine-tipped insects pins allows further insights into individual microvascular patterns of topographically clearly defined brain areas.

| D ISC USSION
If we compare our present findings in adult X. laevis with those gained earlier using the same technique in adult B. bufo (Albrecht et al., 1978;Lametschwandtner & Simonsberger, 1975;Lametschwandtner et al., 1976Lametschwandtner et al., , 1977aLametschwandtner et al., , 1977bLametschwandtner et al., , 1977cLametschwandtner et al., , 1978Lametschwandtner et al., , 1979aLametschwandtner et al., ,b,1980b and with those from a series of carefully performed early LM studies on the brain vasculature of adult anurans (Abbie, 1934;Craigie, 1938;Dierickx et al., 1970Dierickx et al., , 1971Dierickx et al., , 1974Gaupp, 1899;Gillilan, 1967;Millard, 1940;Rex, 1893;Roofe, 1935;Sch€ obl, 1882;Socha, 1930), we find that the pipid frog Xenopus owns a greater intra-and interindividual variation in vessel origins, calibers, and courses than reported in Bufonidae and Ranidae. This has to be questioned, as a comparison of results gained by such different techniques as SEM of vascular corrosion casts and stereomicroscopic analyses of Indi-ink or India-ink/gelatin injected brains with or without prior clearing according to Spalteholz (1914) is problematic. With the latter technique, it is difficult to follow vessel   (Hughes et al., 2014;Saltman, Barakat, Bryant, Brodovskaya, & Whited, 2017) overcomes many of these difficulties. However, it still is difficult to expose individual vessels in both, India-ink or DiI injected specimens over longer distances without destroying vessels, surrounding tissues or anatomical landmarks.
A comparison of the vascular architecture of the brain parenchyma of urodelians (T. cristatus and T. carnifex, A. mexicanum) gained by SEM of vascular corrosion casts (Lazzari et al., 1991;Lazzari & Franceschini, 2003) with that of anurans shows convincingly that urodelians supply the brain parenchyma by hairpin-like vascular loops which arise from the leptomeningeal surface arteries in an acute angle, extend into varying depths of the brain parenchyma before they return in close contact with each other to drain into meningeal veins. Capillary loops are described either as variously bent, inclined or twisted (Lazzari et al., 1991). This pattern contrasts to the anuran pattern where parenchymal vessels form a 3D network, which by its frequent interconnections, clearly enables a better blood supply to the brain parenchyma. Neither in adult B. bufo nor in adult X. laevis, hairpin-like capillary loops are found. Instead, capillaries form a subependymal capillary network which in vascular corrosion casts clearly outlines the contours of brain ventricles.
According to Abbie (1934), the internal carotid artery in R. temporaria divides into cranial and caudal divisions. The anterior division (ramus anterior of Gaupp, 1899) divides at the anterior border of the optic tract into medial and lateral olfactory arteries. Confusingly, a few lines later, Abbie (1934) states that this artery ". . .takes its origin from  It gives off many fairly large branches to the hemisphere (Abbie, 1934).
In R. pipiens, Rana clamitans, and R. catesbeiana, Gillilan (1967) describes also a lateral olfactory artery which gives off a lateral striatal artery and continues as posterior telencephalic artery to supply dorsomedial areas of the telencephalon, epiphysis and thalamus. Interestingly, Craigie (1938) in his comprehensive study on the blood vessels of the brain substance in some amphibians does not describe a lateral olfactory artery in R. pipiens. In Rana esculenta, Socha (1930) does not find a lateral olfactory artery or an artery which takes a similar course. He, however, describes an artery, which ascends between the telencephalic hemispheres and supplies by small branches telencephalic areas. This artery most probably is the ramus hemispherii medialis dorsalis described in R. temporaria by Dierickx et al. (1974), in B. bufo by Albrecht et al. (1978), Lametschwandtner and Simonsberger (1975), Lametschwandtner et al. (1976Lametschwandtner et al. ( , 1977aLametschwandtner et al. ( , 1977bLametschwandtner et al. ( , 1977cLametschwandtner et al. ( , 1979aLametschwandtner et al. ( , 1979b and in X. laevis (this study).
In respect to the lateral olfactory artery described by Abbie (1934) in R. temporaria and by Gillilan (1967) in R. pipiens, R. clamitans, and R. catesbeiana, but not described by Socha (1930) in R. temporaria and by Craigie (1938) in R. pipiens, it is interesting to compare course and pattern of this artery (shown in Abbie's fig. 3) with that of the lateral telencephalic vein found in X. laevis (this study, see Figures 12 and 36). To finally clarify the discrepancies in respect to the presence or absence of this artery in ranid species a SEM analysis of brain vascular corrosion casts of the ranid species cited needs to be done.
In his comprehensive study on the blood vessels of the brain substance in some amphibians, Craigie (1938) mentions that from the transverse anastomotic channel (retroinfundibular communicating artery; Cruz, 1959) ". . .a pair of relatively large arteries run directly dorsal within the brain tissue and branch to supply much of the midbrain and the caudal diencephalon." Our SEM analyses of vascular corrosion casts of the brain of adult X. laevis confirm the presence of these vessels and-due to the advantages of SEM of vascular corrosion casts over binocular dissections or LM analyses of India-ink injected, cleared, and sectioned brain tissue-demonstrates precisely and reliably course, branching pattern, and areas of supply of these vessels.
In adult Xenopus (present study), a closed arterial circle of Willis was found in about two-third of the specimens. In these cases, the circle  (Figures 6 and 7). For a long time, the arterial circle of Willis which was reported to be complete in only 21% of humans studied so far (Lippert & Pabst, 1985) was solely considered to be a compensatory mechanism which in cases of occlusion or stenosis of an internal carotid artery or a vertebral artery enables the redistribution of blood flow (Nornes, 1973). Recently, however, the arterial circle of Willis was considered from an evolutionary point of view and an additional function, namely a function as a passive pressure dissipating system which protects cerebral arteries and the blood-brain-barrier from hemodynamic stress was postulated (Vrselja, Brkic, Mrdenovic, Radic, & Curic, 2014). In men, the arterial circle of Willis comprises anterior and posterior communicating arteries which have much thinner calibers (0.3 mm) than the internal carotid arteries or basilar artery, whereas the arteries forming the arterial circle in Xenopus have rather similar calibers (see Figures 1 and 5). It therefore remains open if the arterial circle of Willis in Xenopus serves the same two FIG URE 46 Microvascular anatomy of the accessory olfactory bulb (aob). Ventro-lateral view. Note the dense capillary bed supplied by a branch of the lateral olfactory artery (loa) and the drainage by ventral, lateral, and dorsal branches of the lateral telencephalic vein (ltv) FIG URE 47 Microvascular pattern of the cochlear-vestibular complex as seen from the rhombencephalic fossa. By the parasagittal section part of the subependymal capillary bed is sectioned-off and underlying supplying arterioles are seen. Note that several capillaries join into single venules which ascend parallel to drain into the choroid plexus of the fourth ventricle (cp IV) functions as that in men where additionally much higher blood pressures in combination with the rigid cranial cavity exert a much higher pulsatile stress to brain vessels (Vrselja et al., 2014). If the higher prevalence of a closed arterial circle of Willis in X. laevis, a species which is secondary aquatic (i.e., derives from a terrestrial ancestor species), is the heritage from the initial land-living lifestyle or is simple due to the nature of its arterial circle cannot be answered yet. To do so, data on the arterial circle of Willis from other land-living amphibians is needed.
X. laevis has a diencephalic and a rhombencephalic choroid plexus only; it lacks choroid plexuses in the telencephalic (lateral) ventricles. It is assumed that choroid plexuses participate to 80% in the formation of cerebrospinal fluid (CSF) by two means, namely by (a) passive filtration of fluid across the highly permeable capillary endothelium and (b) regulated secretion across the single-layered choroidal epithelium (Brinker, Stopa, Morrison, & Klinge, 2014). In respect to (a) the high venous input-in comparison with the small arterial input found in both choroid plexuses of X. laevis-is remarkable. In studies on structure and function of choroid plexuses in mammals and man, which own two choroid plexuses in the lateral ventricles, and one each in the diencephalon and the rhombencephalon, an inflow via a choroidal artery and an outflow through venules is described (e.g., Meeker, Williams, Killebrew, & Hudson, 2012); an additional venous inflow from surrounding cerebral areas is not reported.
Interestingly, vascular casts of choroid plexuses III and IV show many holes of different shapes and sizes indicative of ongoing intussusceptive (nonsprouting) angiogenesis (Figures 23 and 45). Obviously, intussusceptive angiogenesis, a process by which preexisting vessels split or remodel through the formation of transluminal tissue pillars (Caduff, Fischer, & Burri, 1986;Diaz-Flores et al., 2017;Mentzer & Konerding, 2014) occurs not only during tissue development or in some pathological processes including tumours (Diaz-Flores et al., 2017;Mentzer & Konerding, 2014), but also in adult tissue. Intussusceptive angiogenesis by its facets (intussusceptive microvascular growth, intussusceptive arborisation, intussusceptive branching remodeling, intussusceptive pruning; Burri, Hlushchuk, & Donov, 2004) serves to adapt a preexisting vasculature to changing needs of tissues supply and waste removal. In case of the adult choroid plexuses with their high venous inflow, intussusceptive angiogenesis may serve to optimize vascular perfusion for CSF production.

ACKNOWLEDGMENTS
We thank Dr. W.D. Krautgartner for technical assistance in the SEM facility and Christine Radner for preparing and processing brain tissue sections.