J. Anat. (1991), 178, pp. 1-9 1 With 8 figures Printed in Great Britain Evidence for a direct arteriovenous connection (A-V shunt) between the testicular artery and in the of the tree shrew (Tupaia glis)

WORAWUT RERKAMNUAYCHOKE*, TAKAO NISHIDAt, MASAMICHI KUROHMARU AND YOSHIHIRO HAYASHI Department of Veterinary Anatomy, Faculty of Agriculture, University of Tokyo, Bunkyo-ku, Tokyo 113, Japan (Accepted 20 March 1991)

INTRODUCTION A high concentration of testosterone is essential for spermatogenesis in the testis (Steinberger, 1971; Steinberger, Root, Ficher & Smith, 1973) and for prolonging the life span of the epididymal sperm (Reeves, 1987). It has been demonstrated that in order to maintain high concentrations of testosterone intratesticularly, testosterone has to be transferred from the pampiniform plexus to the testicular artery. This has been shown in the rat (Free, Jaffe, Jain & Gomes, 1973; Free & Jaffe, 1975, 1978), ram (Jacks & Setchell, 1973; Ginther et al. 1974), rhesus monkey (Dierschke et al. 1975), cow (Amann & Ganjam, 1976) and man (Bayard, Boulard, Huc & Pontonnier, 1975). The structure of the vasculature in this area has been modified to some extent for this purpose. Although there is no doubt that a countercurrent system is present in this area, there is disagreement as to the way testosterone is transferred into the arterial blood. Some reports- have suggested that it is achieved through an arteriovenous connection (A-V shunt) in the cord (Godinho & Setchell, 1975; Noordhuizen-Stassen, Charbon, de Jong & Wensing, 1985). However, no morphological evidence for the existence of a shunt has been presented. Alternatively, the suggestion has been made that testosterone is transferred by diffusion (Free et al. 1973; Free, 1977). In this study, the first evidence for an A-V shunt between the testicular artery and the pampiniform plexus in the tree shrew spermatic cord is described by light and electron microscopy. The ultrastructural features and functional significance of the A-V shunts are discussed.

MATERIALS AND METHODS Eight adult tree shrews (Tupaia glis) bought from the market in Thailand, weighing 120-170 g, were used in this study. Under ether and pentobarbital anaesthesia (5 mg/100 g body weight, i.p.), the thoracic aorta was cannulated just above the diaphragm. Ringer solution was perfused through the cannula followed by fixative, either 2 5 % glutaraldehyde in 0-1 M phosphate buffer for electron microscopy, or Bouin's fluid and/or 10% formalin for light microscopy.

* To whom correspondence should be addressed. t Present address: Department of Animal Husbandry, College of Agriculture and Veterinary Medicine, Nihon University, Fujisawa 252, Japan. 2 W. RERKAMNUAYCHOKE AND OTHERS

Light microscopy Two paraplast embedded samples were serially sectioned at 6-7 ,tm and stained with Masson's trichrome. Transmission electron microscopy Two samples were excised and cut into smaller blocks (1-2 mm thick), which were subsequently fixed in 2 5 % glutaraldehyde in 0 1 M phosphate buffer for 2 h. They were rinsed in the same phosphate buffer and postfixed in 1 % osmium tetroxide for 2 h. The specimens were then dehydrated in a graded series of ethanol and embedded in Epon 812 and Araldite 6005. Ultrathin sections were stained with uranyl acetate and lead citrate and observed with a JEM-1200EX transmission electron microscope at 80 kV. Scanning electron microscopy For the preparation of corrosion casts, methacrylic methyl ester monomer (Acry ester M., Mitsubishi Rayon Co. Ltd, Japan) was semipolymerised according to the method described by Murakami (1970, 1975). The semipolymerised ester was mixed with 25-30% hydroxypropyl methacrylate, 0-1 % Sudan III and 1-15% benzoyl peroxide (catalyst). Thereafter the mixture was supplemented with 1-1 5% N-N- dimethyl aniline (accelerator) just before injection. The resin was then perfused through the same cannula as described above. The resin-injected samples were immersed in warm water (60 °C) for 4 h, macerated in warm 2-5 % sodium hydroxide for 12-20 h, and then washed in running water followed by distilled water for 1-2 h. Finally, the casts were air-dried, mounted on stubs, and coated with platinum- palladium to be examined with a Hitachi S-4000 scanning electron microscope at 5-10 kV.

RESULTS Testicular artery and pampiniform plexus The testicular artery arose from the abdominal aorta slightly below the origin of the renal artery. It was a small muscular type of artery consisting of 3-4 layers of smooth muscle cells in the tunica media. The testicular (pampiniform plexus), on the other hand, was composed of a single smooth muscle layer in the tunica media and shared a common tunica adventitia with the apposed artery. In the initial portion, the artery ran straight, accompanied by a single (Fig. 1). At about one-third of the distance from the origin, the artery began to become convoluted. The degree of convolution gradually increased as the artery ran caudally to the testis and was prominent in the middle and distal portions of the cord. The vein in this portion also

Fig. 1. LM of paraffin section showing testicular artery (A), vein (V) and capillary networks (*) in the proximal portion of the spermatic cord. The capillary networks are located on the surface of the spermatic cord. Masson's trichrome. x 160. Fig. 2. SEM ofcorrosion cast showing the course ofthe capillary network on the spermatic cord. The capillaries are long, tortuous and twist into U or S shapes (dotted line) and make several anastomoses with each other. Most of the capillaries show dichotomous branching (arrowheads). V, venule. x 670. Fig. 3. TEM of capillary network in the spermatic cord. The network is dominated by capillaries (C) and vesiculated cells. Arrowheads, vesiculated cell vesicles, Co, collagen; Nu, vesiculated cell nucleus. x 4320. Spermatic cord arteriovenous connection 3

I C

,}' F: es rX 7t C 4 .>... i.K.S-s.A.S.... 4. . 4 W. RERKAMNUAYCHOKE AND OTHERS

Fig. 4. SEM of the corrosion cast showing the arteriovenous connection (dotted lines) in the capillary network of the spermatic cord. The connection is formed between a small arteriole (A) derived from the testicular artery and a venule (V) from the testicular vein. x 340. Spermatic cord arteriovenous connection 5 divided into several branches, anastomosing with each other (pampiniform plexus) and covering the artery.

Morphological architecture of A- V shunts Light microscopy and corrosion cast samples Five or six small ramifying arterioles arose directly from the testicular artery at regular intervals. In the initial portion, the tunica media of the arteriole consisted of two smooth muscle cell layers. The muscle layers became thinner as the vessel advanced peripherally. Thereafter, the arteriole gave rise to numerous capillaries. These capillaries were long, tortuous and twisted into U or S shapes in their course (Figs 2, 4). The capillary usually made a series of anastomoses with its counterpart derived from the same arteriole (Fig. 2). However, it occasionally anastomosed with the capillary derived from a different arteriole to make a connection between capillary networks of different arteriolar origins. Some capillary branches drained into a small venule which was directly connected with the testicular vein (pampiniform plexus). Thus a direct connection between the testicular artery and the pampiniform plexus was formed (Fig. 4). The capillary close to the venule had a less dilated lumen as compared with the remaining capillaries. Although most of the capillaries showed dichotomous branching (Fig. 2), trichotomous or tetramerous branching was also observed. The network was arranged in a localised mass and was situated on the surface of the spermatic cord (Figs 1, 4). Only a few portions of the network penetrated deep into the spermatic cord. In the initial portion of the cord, the capillary network was small and simple and A-V shunts were rare, compared with the middle and distal portions. Ultrastructure of the capillary network and its morphology The capillary network contained three kinds of cell, namely, endothelial cells, pericytes and vesiculated cells (Figs 3, 5). The name 'vesiculated cell' was adopted because of the presence of numerous cytoplasmic vesicles (Fig. 7). The endothelial cells with oval nuclei formed a continuous layer, whereas the pericyte layer was discontinuous. The pericytes, which contained microfilaments and possessed long processes within which were vesicles, were surrounded by the basal lamina of the endothelium and made direct contact with the epithelial cells (Fig. 5). The vesiculated cells which were large and irregular in shape consisted of cell bodies and cellular processes (Figs 3, 6). Because of their complicated cell processes, it was difficult to locate the boundaries of adjoining vesiculated cells. In the cell bodies, an abundance of mitochondria, rough endoplasmic reticulum, a well developed Golgi complex, large vesicles and numerous caveolae connecting with the plasma membrane were observed in the cytoplasm (Figs 6-8). The large vesicles varied in size and occupied almost the entire area of the cell body. Smaller vesicles were dispersed throughout the cell body and the cell processes (Figs 3, 7). The vesicles contained a fluid-like substance or lamellar structures, which seemed to be released (Fig. 7). The nucleus with heterochromatin showed invagination of the nuclear membrane. The heterochromatin of the vesiculated cell nucleus was less well developed than the euchromatin, differing from that of the endothelium and pericytic cells (Fig. 3). In the cell processes, a few mitochondria, small vesicles and numerous caveolae were recognisable. The elongated processes made contact with the adjoining cells and surrounded the capillary. In the areas where pericytes were absent, the cell processes were situated close to the capillary endothelium (Fig. 6). Although contact was 6 W. RERKAMNUAYCHOKE AND OTHERS Spermatic cord arteriovenous connection 7 observed between neighbouring vesiculated cells, no junctions were present between the vesiculated cells and the capillary endothelial cells. Abundant collagen fibres, some macrophages and occasional fibroblasts were found in the tunica adventitia of the capillary.

DISCUSSION There is controversy concerning the presence of an arteriovenous connection (A-V shunt) between the testicular artery and the pampiniform plexus in the spermatic cord. After ligation of the distal testicular artery on the testicular surface, the coloured solution (such as Indian ink) which was infused through the proximal testicular artery was visible in the testicular vein. By using this type of experiment, the presence of A-V shunts has been claimed in the ram (Noordhuizen-Stassen et al. 1985), bull, goat, boar and dog (Wensing & Dijkstra, 1981). However, morphological evidence for A-V shunts was not clearly demonstrated. Noordhuizen-Stassen et al. (1985) suggested that the shunt opened or closed depending on the physiological condition and this property made it difficult for it to be observed. Although Hees, Leiser, Kohler & Wrobel (1984) described the existence of A-V shunts in the bull between the small branches of the testicular artery and the pampiniform plexus by light microscopy on serial sections, they failed to establish the presence of shunts by corrosion casts of SEM samples. In contrast, Amselgruber & Sinowatz (1987) reported that there were no A-V shunts in the bull spermatic cord. In the present study, we have demonstrated the existence of direct A-V shunts between the testicular artery and pampiniform plexus by light microscopy and TEM and by the corrosion cast technique combined with SEM. Small arterioles arose from the testicular artery and gave off several capillaries. The capillaries anastomosed with each other to form a network. Capillaries leaving the network drained directly into the pampiniform plexus to form an A-V shunt between the testicular artery and the pampiniform plexus. So far as we know, this kind ofA-V shunt has not been reported before in this area. Ohtsuka (1984) reported a capillary network located between the testicular artery and the pampiniform plexus in the rat spermatic cord. The capillaries, derived from epididymal arterial branches, were connected with the pampiniform plexus and were regarded as vasa vasorum of the testicular artery. The capillary networks in the tree shrew spermatic cord were large and most of them were located on the surface of the cord. In addition, the testicular artery was covered by the pampiniform plexus and consequently only a few parts of the arterial surface could be seen from the surface of the cord. There is therefore no doubt that the capillary network of the tree shrew does not represent vasa vasorum for the artery and differs from that reported in the rat. From transmission electron microscopic observations, it is clear that the vesiculated cell is present in the

Fig. 5. High magnification of TEM showing capillary endothelium (E) and pericyte (Pe). A pericyte process contains microfilaments and makes contact with the endothelium. C, capillary lumen. x 35 500. Fig. 6. The extent of vesiculated cells in the capillary network is not clearly delimited. A long cellular process (arrowheads) surrounds a capillary (C). Numerous mitochondria (M) are present in its cytoplasm. x 7100. Fig. 7. The nucleus of a vesiculated cell has less heterochromatin than euchromatin as compared with nuclei of the capillary endothelium (C). The vesicles of the vesiculated cell vary in size. There is a single large vesicle (*) and numerous smaller vesicles are dispersed through the cytoplasm (arrowheads). x 2100. Fig. 8. High magnification of TEM showing well-developed Golgi complex (G) in a vesiculated cell. *, vesicle; arrowheads, caveolae. M, mitochondria; Nu, nucleus. x 21 300. 8 W. RERKAMNUAYCHOKE AND OTHERS capillary network of the tree shrew spermatic cord, but is not present in the rat. However, we found no modified smooth muscle cells, as described in A-V shunts in other organs (Molyneux & Bryden, 1981). The origin and exact function of the vesiculated cell are still unknown. A countercurrent system occurs between the testicular artery and the pampiniform plexus. The testicular artery and pampiniform plexus modify their structure in some degree to suit the countercurrent system in this area and vary slightly between species. Some of these modified structures, such as the convoluted arteries and reduced vascular walls achieved by sharing a common tunica adventitia, are generally found in mammals with testes located in a scrotum. The other structural modifications, such as the presence of mast cells in the common adventitial layer in the golden hamster (Rerkamnuaychoke, Nishida, Kurohmaru & Hayashi, 1989) and the presence of a venous portal system in the bull and boar (Hees et al. 1984; Amselgruber & Sinowatz, 1987; Rerkamnuaychoke, Nishida, Kurohmaru & Hayashi, 1991), are considered to be species-dependent modifications. These modifications would make the counter- current transfer in this area more effective. In the spermatic cord of the tree shrew, the modified structures were generally similar to those of the golden hamster except for the rare existence of mast cells. However, the presence of an A-V shunt is thought to be a species-specific modification of the tree shrew. This does not prove that the transfer of substances occurs only by this shunt. In addition, the shunt may control the volume of blood draining into the testis. The nucleus of a cell that is active in protein synthesis will be irregular in contour. This allows for an increase in the surface area to meet the demand for interactions between the nucleus and cytoplasm (Fawcett, 1981; Ghadially, 1984). In addition, a cell with little demonstrable chromatin is usually considered to be more active in protein synthesis (Fawcett, 1981). The nuclei of the vesiculated cells are irregular in shape and have invaginations and the cytoplasm of these cells contains rough endoplasmic reticulum, numerous mitochondria and a Golgi complex. It is therefore suggested that the vesiculated cells are active in synthesising the substance which affects the capillaries. In addition, the junction between adjoining cell processes will reinforce the rigidity of the vascular wall and prevent the capillary from collapsing when the shunt is closed. However, further study is needed to define the exact functions of the cell. SUMMARY In the tree shrew Tupaia glis, 5 or 6 small ramifying arterioles arose directly from the testicular artery and then gave off numerous small capillaries. The capillaries made a series of anastomoses with neighbouring counterpart capillaries to become a complicated network. Some of the capillaries drained into a small venule, which was connected directly with the testicular vein (pampiniform plexus), to form an arteriovenous connection (A-V shunt) between the testicular artery and the pampiniform plexus. This A-V shunt appears to make the transfer of substances from the pampiniform plexus to the testicular artery more efficient. In addition, the shunt may control the volume of the blood draining into the testis. The capillaries were covered by vesiculated cells which were located adjacent to the pericytes. The vesiculated cells contained abundant mitochondria, rough endoplasmic reticulum, a well developed Golgi complex and cytoplasmic vesicles. Their cellular processes were long and surrounded more than one capillary. The morphological features of the vesiculated cells suggest that they may synthesise substances that are released into the network and which affect the activity of the capillaries. Since the Spermatic cord arteriovenous connection 9 cellular processes contacted each other, the cells could provide support for the capillaries and prevent their collapse when the shunts are closed.

REFERENCES AMANN, R. P. & GANJAM, V. K. (1976). Steroid production by the bovine testis and steroid transfer across the pampiniform plexus. Biology of Reproduction 15, 695-703. AMSELGRUBER, W. & SINOWATZ, F. (1987). Zur Beziehung zwischen der Arteria testicularis und den Venen des Plexus pampiniformis beim Bullen. Anatomia histologia embryokogia 16, 363-370. BAYARD, F., BOULARD, P. Y., Huc, A. & PONTONNIER, F. (1975). Arterio-venous transfer of testosterone in the spermatic cord of man. Journal of Clinical Endocrinology and Metabolism 40, 345-346. DIERSCHKE, D. J., WALSH, S. W., MAPLETOFT, R. J., ROBINSON, J. A. & GINTHER, 0. J. (1975). Functional anatomy of the testicular vascular pedicle in the rhesus monkey: evidence for a local testosterone concentrating mechanism. Proceedings ofthe Societyfor Experimental Biology and Medicine 148, 236-242. FAWCETr, D. W. (1981). Nucleus. In The Cell (ed. D. W. Fawcett), pp. 195-302. Philadelphia: W. B. Saunders. FREE, M. J. (1977). Blood supply to the testis and its role in local exchange and transport of hormones in the male reproductive tract. In The Testis, vol. 4 (ed. A. D. Johnson & W. R. Gomes), pp. 39-84. New York: Academic Press. FREE, M. J., JAFFE, R. A., JAIN, S. K. & Gomms, W. R. (1973). Testosterone concentrating mechanism in the reproductive organs of the male rat. Nature New Biology 244, 24-26. FREE, M. J. & JAFFE, R. A. (1975). Dynamics of venous-arterial testosterone transfer in the pampiniform plexus of the rat. Endocrinology 97, 169-177. FREE, M. J. & JAFEn, R. A. (1978). Target organs for testosterone transferred from vein to artery in the pampiniform plexus: the . Biology ofReproduction 18, 639-642. GHADIALLY, F. N. (1984). Nucleus. In Ultrastructural Pathology ofthe Cell and Matrix (ed. F. N. Ghadially), pp. 1-43. London: Butterworths. GINTHER, 0. J., MAPLETOFT, R. J., ZIMMERMAN, N., MECKLEY, P. E. & NuTI, L. (1974). Local increase in testosterone concentration in the testicular artery in rams. Journal ofAnimal Science 38, 835-837. GODINHO, H. P. & SETCHELL, B. P. (1975). Total and capillary blood flow through the testes of anaesthetized rams. Journal of Physiology 251, 19P-20P. HEES, H., LEISER, R., KOHLER, T. & WROBEL, K. H. (1984). Vascular morphology of the bovine spermatic cord and testis. I. Light- and scanning electron-microscopic studies on the testicular artery and pampiniform plexus. Cell and Tissue Research 237, 31-38. JACKS, F. & SETCHELL, B. P. (1973). A technique for studying the transfer ofsubstances from venous to arterial blood in the spermatic cord of wallabies and rams. Journal of Physiology 233, 17P-18P. MOLYNEUX, G. S. & BRYDEN, M. M. (1981). Comparative aspects of arteriovenous anastomoses. In Progress in Anatomy, vol. 1 (ed. R. J. Harrison & R. L. Holmes), pp. 207-227. Cambridge: Cambridge University Press. MuRAKAMi, T. (1970). Application of the scanning electron microscope to the study of the fine distribution of the blood vessels. Archivum histologicum japonicum 32, 445-454. MURAKAMI, T, (1975). Pliable methacrylate casts of blood vessels: use in a scanning electron microscope study of the microcirculation in rat hypophysis. Archivum histologicum japonicum 38, 151-168. NOORDHUIZEN-STASSEN, E. N., CHARBON, G. A., DE JONG, F. H. & WENSING, C. J. G. (1985). Functional arterio-venous anastomoses between the testicular artery and the pampiniform plexus in the spermatic cord of rams. Journal of Reproduction and Fertility 75, 193-201. OHTSUKA, A. (1984). Microvascular architecture of the pampiniform plexus-testicular artery system in the rat: a scanning electron microscope study of corrosion casts. American Journal of Anatomy 169, 285-293. REEVES, J. J. (1987). Endocrinology of reproduction. In Reproduction in Farm Animals (ed. E. S. E. Hafez), pp. 85-106. Philadelphia: Lea & Febiger. RERKAMNUAYCHOKE, W., NISDA, T., KUROHMARU, M. & HAYASHI, Y. (1989). Vascular morphology of the golden hamster spermatic cord. Archives of Histology and Cytology 52, 183-190. RERKAMNUAYCHOKE, W., NISHIDA, T., KURoHMARu, M. & HAYASHI, Y. (1991). Morphological studies on the vascular architecture in the boar spermatic cord. Journal of Veterinary Medical Science 53, 233-239. STEINBERGER, E. (1971). Hormone control of mammalian spermatogenesis. Physiological Reviews 51, 1-22. STEINBERGER, E., ROOT, A., FICHER, M. & SMITH, K. D. (1973). The role of androgen in the initiation of spermatogenesis in man. Journal of Clinical Endocrinology and Metabolism 37, 746-751. WENSING, C. J. C. & DIJKSTRA, G. (1981). The morphological relation between the testicular artery and the pampiniform plexus in the spermatic cord. Acta morphologica neerlando-scandinavica 19, 162.