The Site of Release of the Diuretic Hormone in Rhodnius—A New Neurohaemal System in Insects

Diuresis in freshly fed larvae of Rhodnius is caused by the appearance in the haemolymph of a very active diuretic hormone (Maddrell, 1962, 1963). It has been claimed that this hormone is released from the mesothoracic ganglionic mass (Maddrell, 1963). This paper presents evidence to show that this hormone is in fact released not from the ganglionic mass but from close behind it at a series of swollen axon endings lying at the surface of the abdominal nerves which fan out from the back of the ganglionic mass.

Fifth-stage larvae of Rhodnius prolixus Stål taken from a laboratory culture were used in all the experiments. Nervous tissues were assayed for their diuretic hormone content by grinding them up in Ringer’s solution and adding portions of the resulting breis to preparations of isolated Malpighian tubules (Maddrell, 1963). Material for examination with the electron microscope was fixed in phosphate-buffered glutaraldehyde and, after treatment in osmium tetroxide, the tissue was dehydrated and embedded in Araldite. Thin sections were cut on glass knives using a Huxley microtome and, after double staining with uranyl acetate and Reynold’s lead stain, they were examined in a Philips EM 200 electron microscope.

The evidence which leads one to believe that the diuretic hormone is released from the surface of the abdominal nerves close to the mesothoracic ganglionic mass comes from three sources.

T0 expose the abdominal nerves (see Text-fig. 2) insects were opened from the dorsal side under Ringer’s solution. The abdominal nerves were removed and assayed for their content of material possessing diuretic activity. They were found to contain a large amount of such material, often as much as 20% of the total activity extractable from the mesothoracic ganglionic mass in which lie the neurosecretory cell bodies responsible for the synthesis of the diuretic hormone (Maddrell, 1963). In a further set of experiments the lengths of nerve which lie behind the mesothoracic ganglionic mass but in front of a line across the middle of the 2nd abdominal segment were cut out and assayed. They were compared for their hormonal content with the lengths of nerve that occur behind the level of the 2nd abdominal segment. The results are displayed in Table 1 and they clearly show that material possessing diuretic activity is largely confined to the lengths of abdominal nerve just behind the ganglionic mass. A further localization of the diuretic hormone was revealed by a comparison of the amounts of diuretic activity to be found in each of the different abdominal nerves. More hormone occurs in the nerves supplying abdominal segments 1, 2 and 3 than in the nerves supplying the other abdominal segments (Table 1).

This study necessarily involved something which has apparently not been attempted previously; the tracing, in an almost complete series of sections, of axons running from individual cell bodies. In this case the axons followed were those leaving the neurosecretory cell bodies that lie at the back of the mesothoracic ganglionic mass. Some at least of these cell bodies are known from direct assay to contain the diuretic hormone (Maddrell, 1963). The axons were found to run forward in the ganglionic mass away from the cell bodies. Upon reaching the neuropile, each axon divides into two branches. One branch runs more deeply into the neuropile and appears to end there; quite possibly this is where the neurosecretory cell receives afferent stimulation. The other branch turns and runs back with other axons as a fibre tract which soon leaves the back of the ganglionic mass as an abdominal nerve. Text-fig. 1 shows diagrammatically the course of a neurosecretory axon, and representative sections can be seen in Pl. 1. The tracing of the neurosecretory axons in the abdominal nerves becomes progressively more difficult; the distances involved are much greater and the axons themselves contain progressively fewer electron-dense granules and they change in diameter and in their position in the nerve. Tracing them by serial section was, therefore, no longer possible and the organization of the system was examined by looking at sections taken at numerous positions along the length of the nerves. In this way the neurosecretory axons were followed for some way down the nerves and although it was not possible to say exactly where they ended, from the appearance of the nerves at the level of the 2nd abdominal segment it is most likely that only a few of them run farther into the abdomen than this. However, some at least of the axons were observed to branch (Pls. 2 a, b) and in a few cases such a branch was seen to run to what appears to be a neurosecretory axon ending just under the fibrous nerve sheath (Pls. 2 c, d). Indeed, a very obvious feature of this region is the presence just under the nerve sheath of numerous such neurosecretory axon endings (Pl. 3a). They are characteristically swollen with neurosecretory granules; they contain no neurotubules; and not only is there no process from a Schwann cell intervening between them and the nerve sheath, but the nerve sheath over them is often much thinner than it is elsewhere and may be as thin as 700-800 Å (Pl. 3b, c). The large number of these endings can be judged from the fact that a single transverse section of one nerve can show as many as twenty axon endings. Several of the endings contain groups of vesicles similar in appearance to synaptic vesicles (Pl. 4a). It has been suggested that such vesicles are involved in some way in the release of neurosecretion from neurosecretory axon terminals (for a discussion of this point see Bern & Hagadorn (1965) and Johnson (1966)).

The neurosecretory axons are not uniformly distributed among the abdominal nerves: six axons run into each of the abdominal nerves which supply abdominal segments 1 and 2, three enter each of the nerves which run to abdominal segment 3 and only one goes into each of the other nerves which supply the rest of the abdomen. The occurrence of neurosecretory axon endings follows a similar pattern in that many more are to be found at the surfaces of the nerves which innervate abdominal segments 1, 2 and 3 than in the others. It is not clear whether this is merely a reflexion of the numbers of axons in the nerves or whether it is that only one or a few axons branch in each nerve but that they branch more profusely in the more lateral nerves. Possibly some of these neurosecretory axons which originate from cell bodies in the mesothoracic ganglionic mass do not branch and are the ones which run right out to supply the epidermis in which neurosecretory axon terminals have been found (Maddrell, 1965, 1966). However, it is clear that the majority of neurosecretory axon endings to be found at the surface of the abdominal nerves are confined to those lengths of the more lateral nerves which lie closest to the mesothoracic ganglionic mass.

From this ultrastructural picture one can say that the axons from the neurosecretory cells at the back of the mesothoracic ganglionic mass run out into the abdominal nerves where some of them at least branch and end in the proximal lengths of nerve as swollen structures separated from the haemolymph solely by the nerve sheath.

Since the distribution of axon endings corresponds with that of the diuretic hormone revealed by direct assay (section 1), it seems very likely that large numbers of the axon endings contain the diuretic hormone, especially since some at least of the neurosecretory cell bodies from which the axons come have been directly shown to contain the hormone.

The system looks as if it operates to release diuretic hormone not from the mesothoracic ganglionic mass but from a large number of points spread widely over the surface of the abdominal nerves, especially in the region just behind the mesothoracic ganglionic mass.

The appearance of the neurosecretory axon endings in an insect fed 1 hr. before its nervous system was fixed backs up this idea. The swollen axon endings appeared to contain fewer granules and there was some evidence of release of neurosecretory granules by extrusion (Pl. 4b−d) similar to that which is thought to occur during the release of neurosecretory material in other animals (see, for example, Weiss, 1965; Normann, 1965; Smith & Smith, 1966).

However, the following series of experiments were carried out as a more direct test of this suggestion.

In these experiments freshly fed insects were constricted at two different positions behind the level of the mesothoracic ganglionic mass to see what was the effect on subsequent diuresis. If the idea put forward at the end of section 2 is correct, then a constriction at the level of the 2nd abdominal segment (BB in Text-fig. 2) should stop diuresis because it comes between the release point of most of the hormone and the Malpighian tubules; conversely a ligature just behind the mesothoracic ganglionic mass (AA in Text-fig. 2) should not affect diuresis. The insects were constricted by squeezing them with fine forceps held shut with clamps. At the level of the junction of the mesothorax and metathorax, i.e. just behind the ganglionic mass the constriction used was one in which the blades of the forceps were more or less uniformly separated across the width of the thorax by about 150μ (measured under a dissecting microscope fitted with a micrometer eyepiece). The cuticle of the abdomen is much thinner and more flexible than that of the thorax and so it was feared that a tight constriction here might crush the abdominal nerves. To prevent this a thin glass rod of diameter about 150/4 was held between the tips of the forceps. This resulted in a uniform constriction comparable to the one used on the thorax, though, as explained below, because the cuticle is thinner such a constriction on the abdomen might reasonably be expected to be less effective than one on the thorax. The rate of diuresis in the constricted fed insects was followed by allowing the drops of urine to fall into wax-lined Petri dishes filled with liquid paraffin when their diameters could be measured and a note taken of the time at which they were produced (Maddrell, 1964a). Text-fig. 3 displays some typical results. Clearly, a constriction placed just behind the mesothoracic ganglionic mass does not prevent an effective and long-lasting release of the diuretic hormone into the haemolymph behind the constriction. However, a constriction at the level of the 2nd abdominal segment usually stops diuresis after a short period during which the diuretic hormone already in the haemolymph is being used up. That this is not due to damage to the nerves was shown in a number of cases where diuresis was resumed after the constrictions were removed. These results are obtained in spite of the fact that the cross-sectional area occupied by the haemocoel, through which hormone in the haemolymph might move, must be much less under a constriction to 150 μ on the thorax than under a similar constriction on the abdomen, because the cuticle of the thorax is thicker than that of the abdomen (about 70 and 30 μ respectively), and the thorax is less wide from side to side when constricted than is the abdomen (about 4·0 and 6·5 mm. respectively). This experiment shows that the release of the diuretic hormone is largely confined to that part of the insect behind the mesothoracic ganglionic mass and in front of the 2nd abdominal segment.

Since this conclusion tallies with the suggestion made above from a consideration of the ultrastructure of the system and of the results of the assays for the diuretic hormone, it is difficult to escape the conclusion that the diuretic hormone in Rhodnius is indeed released from the series of swollen neurosecretory axon endings which are found dotted over the surface of the abdominal nerves.

The nature of this system for the release of the diuretic hormone, which constitutes a neurohaemal system, is at first sight a little unexpected in that it involves many release points spread out over the surface of peripheral nerves. It is worth pointing out that very few insect hormones are released in such quantities that they can be detected in a single sample of haemolymph (for a discussion of this point see Fraenkel & Hsiao, 1965) but the diuretic hormone is one (Maddrell, 1962,1963). In fact the system for the release of the diuretic hormone is not an unreasonable one if one considers what sort of system might be involved in releasing such a hormone as the diuretic hormone which has a profound effect (acceleration of Malpighian tubule secretion by more than a thousand times) for a short space of time (a few hours) and which starts to take effect very quickly (within 2−3 min. of feeding starting (Maddrell, 1963)). A large number of release sites as in this system for the release of the diuretic hormone must obviously increase the rate at which an effective concentration of the hormone in the haemolymph can be attained. Depending on whether it is considered that flow of neurosecretory granules down the axons would be fast enough to replenish the axon endings or not, a large number of endings would reduce the speed at which such a movement would have to occur, or would allow a relatively slow rate of release of hormone at each axon ending to add up to a very much higher rate for the system as a whole so that release during a few hours might occur without replenishment of individual axon endings. The thinness of the nerve sheath over some of the axon endings should allow the hormone to appear in the haemolymph very soon after being released from the axon. It seems clear that the system for releasing the diuretic hormone in Rhodnius is rather well adapted to its function.

Such a system is not the only way in which a rapid response can be achieved. A more local delivery of a pharmacologically active substance direct from a peripheral efferent nervous supply such as has been suggested for the plasticization of the abdominal cuticle in larvae of Rhodnius (Maddrell, 1965, 1966) and for the acceleration of the heart rate in Periplaneta (Johnson & Bowers, 1963; Johnson, 1966) could lead to such a result. However, Malpighian tubules are not known to be innervated and their structure is matched with their function in that they expose a large surface area to the haemolymph. It is not surprising, therefore, that they are controlled by a hormone carried in the haemolymph.

A close parallel to the hormone-releasing system described in this paper for Rhodnius is to be found in the pericardial organs of crabs (Cooke, 1964; Maynard & Maynard, 1962). In this organ also there are axon endings packed with neurosecretory granules which are arranged at the surface of peripheral nerves. Here too the system releases a hormone which causes a rapid response, in this case of the heart rate, and finally here too the hormone is sufficiently concentrated in the haemolymph to be detectable in a single sample.

The discovery of this neurohaemal system in Rhodnius, and analogy with the situation in Crustacea, leads one to suggest that there may turn out to be several different sites in the central nervous system of insects other than the corpus cardiacum from which hormones may be released into the haemolymph. Indeed, recent work by Raabe (1965, 1966) and Chalaye (1966) has described structures in the stick insect and locust which are associated with the ventral nerve cord and which appear to be sites of accumulation of large amounts of neurosecretion; possibly, therefore, these structures are neurohaemal organs. The neurosecretion involved was found to be rather unusual in that it does not stain with the usual neurosecretory stains, chrome-haematoxylinphloxin and paraldehyde-fuchsin; it does, however, stain intensely with azan. This unusual staining reaction can now be correlated with the appearance of the neurosecretion as seen with the electron microscope. Instead of the more usual electron dense granules, the axons contain rather larger vesicles whose contents are nearly electron transparent (Brady & Maddrell, 1967). Structures similar to these possible neurohaemal organs of the stick insect and locust have now been found in the cockroach (Brady & Maddrell, 1967). Just what hormones these structures might release is not clear, but it is possible that they release some of the products of the neurosecretory cells to be found in the ganglia of the ventral nerve cord.

Some earlier work on the control of the release of the diuretic hormone (Maddrell, 1964b) needs reconsideration in view of the conclusion reached here that the diuretic hormone is released from axons running a short way into the abdominal nerves. In the earlier work it was claimed that the cutting of the abdominal nerves just behind the mesothoracic ganglionic mass stopped diuresis because it prevented sensory information from reaching the neurosecretory cells in the ganglionic mass. It is now clear that this operation must cut the axons from the neurosecretory cell bodies as indeed was suggested by Núñez (1963). However, it has now been possible to cut the abdominal nerves to segments 3−7 at the level of the 2nd abdominal segment, i.e. behind the points from which most of the diuretic hormone is released, and yet the operation still absolutely prevented any diuresis after feeding. Therefore, it still seems reasonable to suppose that sensory information from the abdomen is necessary for the release of the diuretic hormone, the more so when it is remembered that any of the nerves running from the mesothoracic ganglionic mass, other than the abdominal nerves, can be cut without affecting the diuretic response after feeding (Maddrell, 1964b). It is worth noting at this point that diuresis in Anisotarsus also seems to be controlled by sensory information from the abdomen (Núñez, 1956) and a similar situation has now been shown in the cockroach (Dr. R. R. Mills, personal communication).

It is a pleasure to record my thanks to the Science Research Council and to Gonville and Caius College, Cambridge, for the awards of Research Fellowships, during the tenure of which this work was done.

1(a) A field from a section through the mesothoracic ganglionic mass to show a neurosecretory cell body (nsc) and (top right) a bundle of axons which are running back to leave the ganglionic mass as a peripheral abdominal nerve. The bundle of axons contains several neurosecretory axons (asterisks) one of which (double asterisk) is the axon from the cell body in the field, × 5200.

1(b) A section through a group of four neurosecretory axons shortly after they have left the cell bodies and before they reach the neuropile. Note that their profiles contain many more neurosecretory granules than do the profiles of the neurosecretory axons in 1(a) which are farther from their cell bodies, × 11,500.

(a, b ) Two sections at different levels through the same abdominal nerve to show the branching of a neurosecretory axon (asterisk). In 2(a) the axon is beginning to branch, while in 2(b) the branches are separate.2(a) × 15,500; 2(b) × 21,500.

2(c, d) Two sections through the periphery of an abdominal nerve. In 2(c) a neurosecretory axon (black asterisk) touches a neurosecretory axon ending (white asterisk) but is separated from it by a wall (arrow) In 2(d), which is the next section but one, the cytoplasm of the axon is continuous with that of the ending (arrow). Note that neurotubules are confined to the preterminal portions of the axon, (c, d) × 25,5000.

3(a) A section through an abdominal nerve which supplies one side of abdominal segment 1. Note the large number of neurosecretory axon terminals situated at the periphery of the nerve immediately beneath the nerve sheath, × 7500

3(b, c) Sections through large neurosecretory axon endings to show how thin the nerve sheath can be over such endings; in places it is less than 800 A thick. The section shown in 3(b) is a very thin one so that the neurosecretory granules appear uncharacteristically pale. 3(b) × 22,500; 3(c) × 25,500.

4(a) A section through the edge of an abdominal nerve to show two neurosecretory axon endings which contain structures similar to synaptic vesicles (in the areas close to the asterisks). The section also includes the nucleus of a Schwann cell (scn) and axons embedded in Schwann cell cytoplasm which contains very numerous microtubules (mt), × 31,500.

4(b, c, d) Sections through the edge of a small abdominal nerve of an insect fed 1 hr. prior to the fixation of its nervous system. In several places (arrows) neurosecretory granules appear to have been extruded from the neurosecretory axon endings beneath (asterisks), (b, c, d) all at × 60,500.