J. Cell Sci. 60, 117-135 (1983) \ \ 7Printed in Great Britain © Company of Biologists Limited 1983

FUNCTIONAL DESIGN OF MICROVILLI IN THE

MALPIGHIAN TUBULES OF THE INSECT RHODNIUS

PROLIXUS

TIMOTHY J. BRADLEYDepartment of Developmental and Cell Biology, University of California at Irvine,Irvine, California 92717, U.SA.

SUMMARY

The Malpighian tubules of Rhodnius prolixus are divided into two regions; the upper tubule,which is the site of isosmotic secretion and haemolymph filtration, and the lower tubule where waterand KC1 are resorbed.

In the upper tubule the microvilli are arranged in clumps consisting of several hundred microvillilying closely parallel. The microvillar plasma membranes do not touch but are held ~16nm apartalong the full length of the microvilli. As a consequence, the extracellular space between themicrovilli consists of long narrow channels. A morphometric analysis of extracellular, cytoplasmic,endoplasmic reticular and mitochondrial volume within the clumps was conducted. Using thesecretion rate of the epithelium and the channel dimensions, it was calculated that the meanresidence time for secreted fluid in the intermicrovillar spaces was =0-4 s. In view of our currentknowledge of the physiology and morphology of the upper tubule, it is argued: (1) that osmoticallydriven water passes principally through the cells, not the junctional spaces; and (2) that themicrovillar clumps are a morphological specialization, which serves to maximize solute-watercoupling in the upper tubule.

The microvilli in the lower tubule are free-standing, with no pattern of clumping as in the uppertubule. The axopods are about twice as long as the microvilli (10—14/im) and are found in all regionsof the lower tubule. This is in agreement with the proposal that the motile axopods serve to propeluric acid crystals through the lower tubule. No morphological difference was found between theupper and lower halves of the lower tubule, although the two portions are known to be physiologic-ally distinct.

INTRODUCTION

Many investigators have suggested that the pathway for water movement acrossfluid-transporting epithelia is mainly through the extracellular junctional spacesrather than through the cells. In various vertebrate transporting epithelia, such anextracellular pathway has been proposed on the grounds of hydraulic and osmoticconductances (Fischbarg, Warshavsky & Lim, 1977), isosmotic flow at very lowosmotic concentrations (Hill, 1977), solvent drag of non-transported markers (Berry& Boulpaep, 1975; Whittembury, Verde de Martinez, Linares & Paz-Aliaga, 1981),and visible swelling of extracellular spaces during transport (Spring & Hope, 1978).Similarly, in insect rectal epithelium, fluid resorbed from the rectal lumen is thoughtto flow through the extracellular spaces (Wall, 1971; Gupta, Wall, Oschman & Hall,1980). The conclusions drawn from the above observations by no means receiveuniversal acceptance among epithelial physiologists and it has been proposed that

118 T.J.Bradley

unstirred layers around the epithelium make it difficult or impossible to assess ac-curately membrane and epithelial permeabilities, and pathways of water movement(see review by Diamond, 1979).

August Krogh (1929) proposed that for each physiological phenomenon, thereexists an ideal organism or tissue for the study of that process. If that be true, thensurely the Malpighian tubules of the insect Rhodnius pwlixus should rank as primecandidates for the study of rapid, isosmotic secretion. Rhodnius takes a blood mealequal to ten times its previous body weight, which is followed by a rapid diuresisresulting in a reduction of its new weight by 50 % in 2—3 h. This diuresis translatesinto a tremendous flow of fluid across the four Malpighian tubules responsible forurine production. Surprisingly, although this is the most intense fluid-transportingsystem at present known (Maddrell & Gardiner, 1974), the fluid transported isessentially isosmotic over a wide range of osmolalities of bathing fluid (Maddrell,1969).

The Malpighian tubules of R. pwlixus are divided morphologically into an upperand a lower tubule. The upper tubule is the site of production of a haemolymphultrafiltrate (Wigglesworth, 1931), which is drawn osmotically through theepithelium by the active transport of Na+, K+ and Cl~ (Maddrell, 1969). Wiggles-worth (1931) first pointed out that the cells of the upper tubule are characterizedmorphologically by internal spherical crystals and the close-set luminal microvilli areorganized into a striated border. The intercellular spaces are composed of narrow,non-distensible, smooth septate junctions, making them unlikely candidates for thepathway of water flow.

The lower tubule is specialized for resorptive purposes. Maddrell & Phillips (1975)showed that the urine was reduced in osmolality upon passing through the lowertubule, principally due to the resorption of KC1. More recently, Maddrell (1978)showed that the lower tubule, although seemingly a uniform epithelium, had twofunctionally distinct regions. The upper half of the lower tubule shows the greatestwater permeability and is presumably the site of the uptake of water released by uricacid precipitation, while the lower half shows a high rate of active KC1 resorption anddecreased permeability to water. Morphologically, the cells of the lower tubule differfrom those of the upper in that they lack intracellular crystals and show markeddifferences in the organization of the luminal cell surface. The microvilli are morewidely spaced in the lower tubule, forming a brush border (Wigglesworth, 1931).Interspersed among the microvilli are larger, longer microtubule-containing exten-sions called axopods (Bradley & Satir, 1979). These have been described only fromthe lower half of the lower tubule but their purported role in moving uric acid crystalssuggests that they should be found throughout the lower tubule.

In the present study, these three regions of the tubule (upper tubule, upper half oflower tubule, and lower half of lower tubule), which are known to be functionallydistinct, have been examined for corresponding differences in morphological or-ganization. The unique perspective provided by scanning electron microscopyrevealed morphological specializations in the microvilli of the upper tubule. Theseresults are interpreted with regard to the mechanism of solute—water coupling in the

Microvillar design in Rhodnius 119

epithelium. In the lower tubule, new information on the distribution and size of theaxopods was obtained.

MATERIALS AND METHODS

Dissection procedures and preparation for electron microscopyR. prolixus used for these studies were taken from a colony maintained as described previously

(Bradley & Satir, 1981). All the insects used for dissections were in the fifth instar and had beenstarved for more than 1 month. Dissections were conducted on animals pinned to the wax-coveredbottom of Petri dishes filled with insect Ringer solution: llmM-KCl, 129mM-NaCl, 8-5 mM-MgCl2, 2mM-CaCl2, 34mM-glucose, lOmM-Hepes, pH adjusted to 6-9 with NaOH.

Dissected tubules were fixed for transmission electron microscopy by immersion for 1 h in asolution containing 4% glutaraldehyde, 005M-sodium cacodylate and O-lM-sucrose (pH6-9).Following rinsing in buffer, the tubules were post-fixed for 45 min in a solution of 1 % OsO4,005 M-sodium cacodylate and 0-1 M-sucrose, dehydrated in an ethanol gradient and embedded inEpon 812. Sections were stained with 1 % uranyl acetate in 50 % ethanol followed by Reynolds' leadstain. Micrographs were taken on a Siemens 1A or JEOL C100 electron microscope.

Scanning electron microscope (SEM) specimens were fixed in the above glutaraldehyde-containing solution for 1 h and then taken through an ethanol gradient followed by an ethanol-to-Freon 113 gradient series. Tubules were critical-point dried (apparatus by Bomar) using Freon 13.

The tubules were subsequently broken or split with razor blades, as appropriate, to reveal variouscellular features, and placed on an SEM stub covered with double-sticky tape. Specimens werecoated with gold/palladium using a Hummer II sputter coater (Technics). Scanning electronmicrographs were taken on a Hitachi S-500.

Morphological measurementsThe Rhodnius prolixus used for this study and for my previous work on Rhodnius (Bradley &

Satir, 1979, 1981) derive from a colony established 6 years ago by Dr Lauren Zarate of the Univer-sity of California, Berkeley, using insects that she collected in Mexico. These animals will be termedthe University of California (UC) strain. Work cited in this paper that was conducted in theCambridge laboratories of Drs Wigglesworth and Maddrell was performed on insects from theCambridge University colony oiR. prolixus. The Cambridge colony was established over 100 yearsago. Comparisons between these two strains indicate that the animals are essentially identical withregard to body length and weight at any given instar, developmental timing, blood meal size, andrate of diuresis (Zwicky & Wigglesworth, 1956; Maddrell, personal communication). Slight butprobably significant differences appear to exist with regard to the lengths and diameters of theMalpighian tubule segments (Maddrell, 1964; and personal communication). In this paper,therefore, morphological and physiological measurements refer to the UC strain. The microvillarwaves described in the upper tubule are seen in the Cambridge strain as well (Maddrell, personalcommunication).

In order to measure tubule length, entire tubules were dissected from fifth instar larvae and placedin drops of Ringer solution under paraffin oil. Using fine forceps, the tubules were pulled out of thedrop under oil and their lengths were measured using an eyepiece micrometer on a Wild 5MAdissection microscope. Surface tension in the drop provides sufficient resistance to straighten thetubule and allow accurate measurement of its length.

Morphometric quantification of microvillar structural parameters was performed using the point-count method as described previously (Bradley & Satir, 1981), with the exception that themicrographs used were of microvilli cut in cross-section and were printed at a final magnificationof X30000.

Measurements of maximum rates of transportFifth instar larvae that had been starved for 1 month were allowed to feed to repletion on a rabbit.

The insects were weighed immediately after the defaecation that follows the cessation of feeding by1 to 2 min, and at half-hourly intervals thereafter. It has been shown that weight loss in R. prolixus

120 T. jf. Bradley

during the 2-3 h postprandial period of rapid diuresis can be fully accounted for by fluid excretion(Maddrell, 1964).

RESULTS

Each of the four Malpighian tubules in a fifth instar R. pivlixus consists of a longtubular epithelium, which is closed at the upper end and opens into the rectum via anampulla at the lower end. The epithelium is classified as simple, meaning that theepithelial wall separating the luminal space from the haemolymph is one cell thick.

The upper tubule

The upper approximately two thirds of the Malpighian tubule in Rliodnius aredifferentiated into a distinct tubular region, the upper tubule. In fifth instar larvaethis region has a length of 29 ± 1 mm (mean ± s.D., n = 5). The whole upper tubulecontains only one cell type (Wigglesworth & Salpeter, 1962). At its closed end thetubule is capped by cells that appear to be identical to those forming the rest of theupper tubule. The cap is formed by cells growing together to seal the end of thetubule, but the pattern by which this occurs is quite variable (Figs 1—4). Light-microscopic examinations indicate that each bulge as seen in SEM is a separate celland the caps generally consist of two to three cells.

The tubule itself is composed of cells, spaced alternately along the tubule, whichbulge outward away from the lumen giving the upper tubule a zig-zag appearance(Figs 1, 3, 5). The cells tend to resemble half-cylinders, which are thickened in theperinuclear region, and are joined to their neighbours by septate junctions to form atubule.

The transition from upper tubule to lower tubule is abrupt. The point of transitioncan be discerned in scanning electron micrographs, because the lower tubule is smal-ler in diameter and the outlines of individual cells are less prominent (Fig. 5). Theexact point of transition is even more obvious when the tubules are viewed usingtransmitted light (Fig. 6). Under these conditions the upper tubule cells are opaquedue to their intracellular crystals, while the cells of the lower tubule, which lackintracellular crystals, are clear.

Fig. 7 shows an upper tubule broken across and viewed in the scanning electronmicroscope. Several tracheae can be seen lying on the smooth basal lamina covering thetubule. The interiors of the cells are seen to contain clusters of crystals, each of whichlies in a membrane-bound vacuole. The diameter of the tubule in Fig. 7 is29 /Am. Critical-point drying, which is a necessary preparatory step for scanningelectron microscopy, preserves cellular ultrastructure and surface features, but cancause bulk shrinkage of the specimen (Boyde & Wood, 1969). For this reason tubuledimensions, microvillar length and intermicrovillar spacing were measured onspecimens prepared for transmission electron microscopy (TEM) in which critical-point drying was not used. The tubules have an outer diameter of about 45 ^m, whilethe luminal diameter defined by the tips of the microvilli is 15 fim. These values are inagreement with those we have measured in the light microscope on living Malpighian

Microvillar design in Rhodnius 121

Figs 1—4. Scanning electron micrographs of the tips of Rhodnius Malpighian tubules.Note that many patterns of tip closure exist. In most of these, two to three cells growtogether to form the closed end. Fig. 1, X330; Fig. 2, X560; Fig. 3, X330; Fig. 4, X33O.

Figs 5, 6. The transition from upper tubule to lower tubule is abrupt (arrow). In each ofthese micrographs the upper tubule lies to the left of the arrow, the lower tubule to theright.

Fig. 5. In this scanning electron micrograph the upper tubule can be distinguished bymore-exaggerated bulging cells and a resultant zig-zag pattern to the tubule. X80.

Fig. 6. In transmitted light microscopy the upper tubule cells appear opaque due to thepresence of intracellular crystals. X80.

tubules suspended in insect Ringer solution, indicating that preparation for TEMdoes not appreciably distort the ultrastructural dimensions of the tubules.

Wigglesvvorth & Salpeter (1962) described the microvilli of the upper tubule as astriated border. The microvilli do not form a uniform border but rather are clumpedtogether in a wave-like pattern over the luminal surface (Figs 8, 9). Each 'clump' seenalong the luminal surface is composed of hundreds of microvilli (Fig. 10). Themicrovilli in each clump are pressed together and run parallel to each other alongalmost their entire length (Fig. 12). By examining cross-sections of the microvilli inthese clumps at higher magnification, one can observe that the microvilli are not onlyclosely aligned but the positioning of their plasma membranes is quite specific (Fig.11). At the points where two microvilli are closely adjacent, the plasma membranesof each run parallel, maintaining a rather precise 16nm space between them. This

CEL60

T. y. Bradley

Figs 7-10. Scanning electron micrographs of upper Malpighian tubules.Fig. 7. An upper tubule, broken transversely. The exterior of the cell is covered with

a smooth basal lamella (bl). All the tracheae (t) and tracheolar cells are exterior to thislamella. The cytoplasm of the cell contains clusters of membrane-bound vacuoles (v). Themicrovilli in the lumen (/) are difficult to discern in non-split tubules. X 1730.

Fig. 8. A portion of a tubule broken to reveal both the luminal surface with the clumpsof microvilli (arrows) and the external basal lamella (bl). X3000.

Fig. 9. A tubule split longitudinally to reveal the pattern of microvillar clumping withinthe lumen (/). X1800.

Fig. 10. A higher magnification view of the luminal surface from a preparation similarto that in Fig. 9. Note that each clump is composed of numerous microvilli. X9120.

Miavvillar design in Rhodnius 123

Fig. 11. A transmission electron micrograph of a thin section taken perpendicular to thelong axis of microvilli within a clump in the upper tubule. The plasma membranes ofadjacent microvilli do not touch, but maintain a rather precise ~16nm spacing (arrows).X53 250.

Fig. 12. Lower magnification transmission electron micrographs reveal the pattern ofmicrovillar clumping in the upper tubule if the plane of section is appropriate. Since onlyone cell junction \j) is visible in this micrograph, we can be certain that all three clumpsto the left of it occur in the same cell. X7500.

124 T.J. Bradley

space is slightly electron-dense, possibly because of the glycoproteins present on theouter face of the microvillar plasma membrane. As a consequence of the plasmamembranes from two adjacent microvilli running parallel, the microvilli appear ir-regularly polygonal in cross-section, rather than round.

At present it is not clear what determines the size, shape and border of each clumpof microvilli. The clumps are not fixation artefacts since identical stationary 'wave-like' patterns are visible in living non-fixed material using differential interferencecontrast optics in the light microscope. From both scanning (Fig. 8) and transmissionelectron microscopy (Fig. 12) it is apparent that there are many clumps on the surfaceof each cell. Microvilli, both with and without mitochondria, participate in the pat-tern.

Morphometric analysis of the microvilli of the upper tubule

The Malpighian tubules of R. prolixus have been more intensively studied than anyother insect Malpighian tubule. As such they serve as an important model tissue forinvestigating mechanisms of isosmotic secretion. The most intractable questions havebeen: (1) the cellular locations and biochemical nature of the ion pumps; (2) thepathways of water movement; and (3) the precise manner in which these two interactto yield an isosmotic secretion. Recently, these questions have begun to be addressedusing the tool of mathematical modelling (Maddrell, 1980). For this reason, a precisequantitative description of the organization of the microvillar clumps in the uppertubule would seem warranted.

Morphometric analyses were carried out on sections taken through the microvillarclumps and perpendicular to the long axes of the microvilli, as in Fig. 12. Measure-ments were made of the percentage volume (Vv) of extracellular space (i.e.intermicrovillar space), cytoplasm, mitochondria and endoplasmic reticulum withinthe clumps (Table 1). Extracellular space occupies only about one-fifth of the totalvolume, illustrating the very close placement of the microvilli. About half of thevolume within the clumps consists of cytoplasm within the microvilli. The remainingintracellular volume is occupied largely by mitochondria, with the volume ofendoplasmic reticulum being relatively small. Some of the measurements contribut-ing to the mean values given in Table 1 were made on sections taken near the tips ofthe microvilli, others from sections taken near the base. When these were comparedseparately, no significant difference correlated with position along the microvillus was

Table 1. Volume density (Vv) of various ultrastructural compartments within themicrovillar clumps in the upper tubule

Extracellular Cytoplasm of Mitochondria Endoplasmicspace microvilli reticulum

Mean 185 49-6 29-7 2-2S.D. 1-9 2-5 16 0-9

w = 4, with four determinations on each of four tubules.

Microvillar design in Rhodnius 125

observed for any of the morphological compartments. The values in Table 1 representthe means for the various compartments at any point along the microvilli, which arevery uniform along their length.

Maximum rate of fluid transport by the upper tubule

The highest rates of fluid transport by Rhodnius Malpighian tubules occur duringthe rapid diuresis following the blood meal. At this time the upper tubule is secretingmaximally, while the lower tubule is resorbing KC1 but little fluid (Maddrell &Phillips, 1975). The volume of urine produced during this period is therefore essenti-ally identical to the volume of fluid secreted by the upper tubules. This rate isconsidered maximal because, in the intact animal, the tubules are bathed in naturalhaemolymph, are fully oxygenated by tracheae, and are stimulated by a saturatingconcentration of the natural diuretic hormone (Maddrell, 1964).

The rate of urine production was measured in three fifth-instar Rhodnius prolixusat 27 °C, the temperature at which they had been reared. The maximum weight lossesobserved in one half-hour period were 21-8 mg, 22-4 mg and 22-1 mg, respectively, forthe three insects. These values demonstrate the uniformity in the rate of diuresisexhibited by these insects. The largest observed weight loss, 22-4 mg, represents anaverage excretion rate by the insect during that period of 750nl/min. The fourMalpighian tubules in Rhodnius are identical in size, so it is reasonable to assume thatall four secrete at identical rates. The maximum secretion rate per tubule wouldtherefore be one-fourth of the total rate or 187nl/min. This maximum rate in vivois well above the maximum rate achievable in vitro. In our own studies using tubulesisolated in a droplet of Ringer under oil and stimulated with 5 X 10~5 M-5-hydroxy-tryptamine, the maximum observed rate of transport was 60nl/min, a value aboutone-third of that observed in vivo.

The upper tubule closely resembles a cylinder 45 [im in diameter (d) and 29 mm inlength (/). The external surface area of this cylinder, ignoring membrane folding, istherefore ndl or 4-lmm . Dividing the above maximum rate of secretion by thesurface area of the tubule yields a transport rate per unit area of 46nlmin~' mm~2

(=O77jim3s~Vrn~2)-

The lower tubule

The lower one-third of the Malpighian tubules in Rhodnius is differentiated into adistinct tubular region, the lower tubule. In fifth instars the length of this region is13-5 ± 1-6 mm (mean ± s.D., n = 5).

Based on ultrastructural parameters, the lower tubule of Rhodnius has been con-sidered a uniform epithelium containing only one cell type (Wigglesworth, 1931;Wigglesworth & Salpeter, 1962). Maddrell (1978) showed that the upper and lowerhalves of the lower tubule differed in their physiological characteristics, the formerbeing more water-permeable and the latter showing more rapid K+ transport. Theupper half of the lower tubule was separated from the lower half in all of our prepara-tions and each region was examined separately by both scanning and transmissionelectron microscopy. No morphological or ultrastructural difference between these

126 T.J.Bradley

two regions of the lower tubule was discernible. For this reason, the ensuing descrip-tion of the lower tubule applies to cells along its entire length.

The abrupt change in cell type that occurs at the junction of the upper and lowertubules (Fig. 6) is associated with a concomitant change in the ultrastructure of thecells as observed in the scanning electron microscope. In the lower tubule, nointracellular crystals are seen (Fig. 13) but crystals of uric acid are observed in thelumen (Figs 13, 15). The microvilli in this region are not grouped into clumps (Figs13, 16) but, instead, form a uniform field of individual, relatively erect microvilli.These give the impression of being rather pliable in vivo, since they can be pushedaside in regions where the uric acid crystals nestle among them (Fig. 15).

In a transmission electron microscopic study of the lower tubule, Bradley & Satir(1979) reported that the luminal surface of the cells possessed conventional insectmicrovilli as well as larger, microtubule-containing axopods. Although the length ofthe axopods could not be measured directly, because they bent out of the plane ofsection, our impression from that study was that the axopods were at least 10 pirn long.The present observations using scanning electron microscopy do reveal aheterogeneity in length among the elongate structures on the luminal surface, withsome being perhaps twice as long as others (Figs 13, 14). This would suggest anoverall length of 10-14 fim for the axopods. These longer structures are particularlyprevalent near the cell junctions. Fig. 14 shows a region of tubule in which an indenta-tion is observed that corresponds to the junctional region between two cells. Theluminal surface at the area of the junction shows numerous longer structures. Closeexamination of similar junctional and non-junctional areas, using transmissionelectron microscopy of thin sections, revealed that, while axopods are dispersed overthe luminal cell surface, they are more closely spaced and prevalent near the celljunction (Fig. 17). It is clear from these results that, while scanning electron micro-scopy is excellent for determining the size, distribution and spatial arrangement ofmicrovilli, it is inadequate for differentiating axopods and microvilli unambiguously.For example, in Figs 15 and 16, structures of various forms and lengths are evident,but it is impossible to state with certainty which are axopods. For this purposetransmission electron microscopy of thin sections is necessary in order to identify theinternal microtubules that are diagnostic of the axopods.

Figs 13—16. Scanning electron micrographs of the lower Malpighian tubule.Fig. 13. A lower tubule broken transversely. The exterior of the tubule is covered with a

smooth basal lamella (bl). The cytoplasm of the cell does not contain spherical crystals (com-pare with Fig. 7), but round uric acid crystals (arrow) are observed in the lumen (/). X 1820.

Fig. 14. In this view of a tubule split longitudinally we can see both the external basallamella (bl) and the luminal surface (/). Longer luminal extensions (purported axopods,arrow) are seen lying near the cell junction, which is revealed as an indentation(arrowhead) on the exterior of the tubule. X1700.

Fig. 15. A view of the microvilli covering the luminal surface of the lower tubule. Notethat the microvilli are not clumped. The depressions in the microvillar mat (arrow) werepresumably formed by uric acid crystals such as the one seen in the lower right (c). X 4300.

Fig. 16. At higher magnification it is apparent that the lumen is covered with in-dividual, erect microvilli and axopods. They are not unambiguously distinguishable in thescanning electron microscope (see text). X19250.

Micmvillar design in Rhodnius 127

Figs 13-16

128 T. J. Bradley

. A.tf

Fig. 17. A transmission electron micrograph of a portion of the upper one half of the lowertubule. Large axopods containing the characteristic microtubules (arrowheads) can beseen arising from the luminal surface of the cell near the cell junction (j). The smallermicrovilli, which lack microtubules, are seen both in longitudinal and cross-section(arrows).

DISCUSSION

Wigglesworth (1931) conducted a histological study of the upper and lower Mal-pighian tubules of Rhodnius. He described the upper tubule as composed of a singlecell type characterized by intracellular crystals, lying in membrane-boundedvacuoles, and close-set microvilli organized in a 'striated border'. Later, using theelectron microscope, Wigglesworth & Salpeter (1962) were able to discern that themicrovilli were closely packed, with a space of about 15-20 nm between them. In thepresent study, 1 have extended these observations using the scanning electronmicroscope and have revealed a novel pattern of organization in these microvilli. Inthe upper tubule, the microvilli are arranged in clumps consisting of several hundredmicrovilli lying closely parallel. This pattern of microvillar clumping covers the entireluminal surface of the upper tubule and many clumps are present per cell. Thisarrangement was not observed previously, because the long-range repetition ofclumps is difficult to discern in sectioned material. Electron micrographs of theluminal region taken at low magnification (Fig. 12) can, however, reveal the clumpedpattern if the angle of section is appropriate. For reasons discussed below, the greatest

Microvillar design in Rhodnius 129

functional significance of this arrangement probably lies in the closely parallelpositioning of the microvillar membranes. Thin sections cut perpendicular to the longaxis of the microvilli demonstrate that the clumps consist of microvilli held a ratherprecise distance apart (=16 nm) over most of their length. As a consequence of thisarrangement, the extracellular space between the microvilli consists of long, narrowchannels whose length: radius ratio equals 7000nm/8nm, or 875.

1 am unaware of any other report in the literature of microvillar associations suchas are described here for the upper tubule. It can be demonstrated that the microvillarclumping is not a result of fixation (e.g. cross-linking by glutaraldehyde) since: (1)the microvilli in the lower tubule, fixed under identical conditions, are normal andfree-standing; and (2) fresh, non-fixed material viewed by differential interferencecontrast light microscopy reveals an identical pattern of clumps. Light microscopy ofliving material also reveals that the clumped microvilli are stationary and are thereforenot 'frozen' images of group movements such as are seen in fixed metachronal wavesof cilia (Satir, 1963).

The epithelium of the Malpighian tubules shows an abrupt transition in cell typeat the junction of the upper and lower tubule. In the scanning electron microscope,the pattern of microvillar organization in the lower tubule is seen to be distinctlydifferent from that in the upper. Long, individual cellular extensions are seen, withno pattern of clumping or coordinated bending. The extensions are heterogeneouswith respect to length. Transmission and scanning electron microscopy of similarregions indicates that the larger structures are axopods and that these tend to be onlyabout twice as long as the microvilli. Unequivocal identification of axopods,therefore, depends on identifying microtubules inside the cellular extensions, usingtransmission electron microscopy.

Bradley & Satir (1979) restricted their investigations of axopods to material takenfrom the lower half of the lower tubule. However, since axopods are thought to beatactively, thereby moving uric acid crystals down the tubules, and since the crystalsare present in the entire length of the lumen of the lower tubule, these authors felt thataxopods should also be present throughout the lower tubule. Fig. 17, which is amicrograph of a region in the upper half of the lower tubule, clearly contains axopods.This indicates that axopods are indeed found in both the upper and lower halves ofthe lower tubule. It is interesting that we still have no evidence of morphologicaldistinctions between the upper and lower halves of the lower tubule, although weknow they display physiological differences (Maddrell, 1978). The axopods appear tobe somewhat more common on the luminal cell surface near the cell junctions thanthey are near the centre of the cell. No functional explanation for this arrangementcomes to mind and it may reflect some aspect of the arrangement of the cytoskeletonwithin the cells, rather than the axopods' function in moving uric acid crystals downthe lumen.

Pathways of water movement in the upper tubule

It seems appropriate to consider the clumping pattern observed in the microvilli ofthe upper tubule with regard to microvillar function and the rapid isosmotic flow that

130 T.J.Bradley

occurs in this region. It is generally accepted, mainly on the grounds ofelectrophysiological data, that ion transport in Malpighian tubules occurs by meansof transport 'pumps' located in plasma membranes forming the basal infolds and apicalmicrovilli (Maddrell, 1971; Harvey, 1980). Similarly, there is a general consensus(see Maddrell, 1980; Phillips, 1982; Bradley, 1983) that fluid transport by Mal-pighian tubules is driven by an osmotic gradient. A direct demonstration that osmosisis the driving force for fluid movement in Malpighian tubules is not yet technicallypossible. The indirect evidence can be summarized as follows: (1) the fluid producedin the Malpighian tubules is either isosmotic or slightly hyperosmotic (never morethan 1—2 %) to the bathing medium, even if fluid secretion rates are varied by severalorders of magnitude or if osmotic concentration is varied with non-transported solute(e.g. sucrose); (2) varying individual ion-secretory rates, e.g. by enriching for K+ orsubstituting for Cl~ does not change the osmotic relationship between bathing andsecreted fluid; and (3) at any given level of stimulation, tubules will secrete faster indilute than in concentrated solutions (Maddrell, 1969, 1971). Additional evidencethat fluid production in insect Malpighian tubules involves osmotic coupling waspresented by Maddrell (1980).

The location, at the subcellular level, of this osmotic gradient and the pathways ofwater movement into the lumen are much less clear. Diamond & Bossert (1968)suggested that membrane folds and extracellular spaces might be sites of osmoticcoupling in rabbit gallbladder and, by analogy, in a number of other ion-transportingepithelia. Berridge & Oschman (1969) extended this model to Malpighian tubules,suggesting that the basal folds and spaces between the apical microvilli could serve assites for local osmotic gradients. In such a model, most of the osmotically driven waterwould move through the cell. The observation that large molecules such as inulin(5000 molecular weight) can pass through the epithelium, in the apparent absence ofvacuolar transport, has promoted the idea that some proportion of the transportedsolutes must pass through the septate junctions between cells, since diffusion acrosscell membranes seems unlikely for such large compounds (Maddrell & Gardiner,1974). The mechanism for moving compounds through the junctional space, whichis of the order of 5-7 pm in length, would presumably be solvent drag if it were a rapidprocess and diffusion if it were a slow one. More recently, Maddrell (1980) haspresented the results of preliminary studies on the rate of inulin movement throughthe upper tubule of Rhodnius. He compared the rates of inulin appearance in the urineduring slow and rapid secretion and found that the rate of inulin movement into thelumen was unaffected by fluid secretion rate. Since inulin is thought to pass throughthe epithelium via the junctional spaces, Maddrell's results argue that the greater partof transepithelial water-flow during diuresis does not occur through the extracellularpathway but rather through the cell. Our observations support this view, since thereare no ultrastructural specializations associated with the apical ends of the junctionsthat would allow water to be drawn osmotically through the junctions at the requiredrate.

Maddrell & Phillips (1975) pointed out that the organization of the microvilli inRhodnius Malpighian tubules is related to the osmolality of the fluid being transported.

Microvillar design in Rhodnius 131

The microvilli in the upper tubule are close-set and the transported fluid is isosmotic.The microvilli in the lower tubule are more widely spaced and the resorbed fluid inthat region has an osmotic concentration of about 1 -7 OSM. On the basis of the resultsof the present study, it is now clear that the microvilli in the upper tubule are notmerely closely spaced but are organized into clumps. As a result, the plasma mem-branes form long, narrow extracellular channels in which solute-water coupling ismaximized. We believe that the microvillar clumping observed in the upper tubuleis a specific adaptation promoting solute—water coupling in this secretory tissue.

Rates of fluid flow in the intermicrovillar spaces

In Diamond & Bossert's (1968) standing osmotic gradient model, ion transport intoopen-ended membranous channels is thought to produce local hyperosmotic compart-ments, which serve to couple solute and water transport across the epithelium. If asimilar model is applied to Malpighian tubules (Berridge & Oschman, 1969), itfollows that the apical hyperosmotic spaces would lie in the intermicrovillar clefts andthat the osmotically driven fluid flow would also occur between the microvilli. Thenarrowness of the spaces between the microvilli in the upper tubules of Rhodniuspromotes isosmotic secretion by reducing the distance over which the equilibratingdiffusion must occur; yet it also speeds fluid flow, a factor that would reducesolute—water coupling. I wished to estimate the rate of fluid flow through theintermicrovillar spaces to determine: (1) whether such a pathway could be ruled outon the basis of unrealistic rates of flow; and (2) whether the residence time within thechannels would appear to be too short (e.g. of the order of milliseconds) to allowosmotic equilibrium within the spaces. Our morphometric measurements of theupper tubule microvilli allow such estimates to be made.

If we assume: (1) that all the fluid flowing through the cell must pass into theintermicrovillar spaces, with no flow through the closed ends of the channels or outthe tips of the microvilli; and (2) that fluid addition occurs at a constant rate along thelength of the channel (see Fig. 18A), then the volume of fluid flowing past any pointin the channel increases linearly with respect to position in the channel:

Vo = kl, (1)

where Vo is volume (/im3s~'), k is a constant with units/imzs-1 and / is the distancefrom the closed end of the channel (jUm). Our morphometric data indicate that thechannels do not change cross-sectional area with length. Therefore, the velocity offlow also increases linearly within the channel:

where Ve is the velocity in /im s"1 and A is the cross-sectional area of the extracellularspaces (channels).

The cells of the upper tubule resemble flattened cups that are slightly longer in thedirection parallel to the long axis of the tubule. The cells cover one side of the tubuleand are staggered in their positions along the tubule (Figs 2, 3). For our purposes

132 T.J.Bradley

here, the cells can be considered to resemble closely one-half of a cylinder. Thetubules average 45 fim in diameter, the lumen 15 fim in diameter. The cells are about95 fim in length parallel to the long axis of the tubule and 71 fim in outer circumferenceperpendicular to the long axis of the tubule. The external or basal surface area of anupper tubule cell, ignoring membrane folding, is therefore 6745 fim2. Multiplying thisvalue by the secretion rate per unit surface area (0'77/im s" ^m ), we arrive at5194^m3s~' as the secretion rate for a single cell. The area of luminal surface on sucha cell, again ignoring membrane folds, is naturally less than that of the outer surface.The calculated surface area for the luminal space approximating a half-cylinder andlying just apical to the tips of the microvilli is 2238 fim2. The morphometric measure-ments obtained in this study indicate that 18-5 % of the cross-sectional area of themicrovillar border is intermicrovillar space (Table 1). The aggregate surface areathrough which the fluid is actually flowing out is therefore 2238jtzm2 (18-5%) =414/im2.

We can calculate the value of k using equation (1). Vo per cell is 5194^m2s~' andthe curved microvilli have a length of about 7 fim. k therefore equals 742 fim s"1. Bysubstituting the above value for the surface area of the intermicrovillar spaces backinto equation (2) we find that the maximal velocity (Ve) of fluid flow, namely that flowat the tips of the microvilli, is:

This calculation allows us to provide units for Fig. 18B, since we know that velocityof flow within the channels varies linearly with length from 0 to 12-5 fim/s. Maddrell(1980) calculated the percentage of the external surface area of the upper tubule ofRhodnius, occupied by the intercellular clefts, and arrived at a value of 0-034%. Bycomparison, the area of intermicrovillar space is 6 % as large as the area of the outsideof the tubule. If all the fluid transported by the upper tubule were to pass through theseptate junctional space it would, therefore, have to travel at a rate 200 times fasterthan would fluid moving through the microvilli.

Solving for dt in equation (2) and integrating gives an expression that can be usedto calculate the transit time for fluid within the channel, i.e.:

A i d /

" J T-The transit time between two points, l\ and li, is given by:

This relationship is shown graphically in Fig. 18c. Since the volume of flow varieslinearly with distance in the channel, the median transit time is equivalent to thetransit time of fluid entering at the midway point, i.e. where h/l\ = 0-5. Solving fort we find that the median transit time for fluid entering the intermicrovillar spaces is0-4 s. This would seem to be a relatively long residence time, but a final determination

Microvillar design in Rhodnius 133

-Length (/)-

I

Fluid velocity atthe microvillar tips

Position in channel (!)

CD

Median residence time

Relative position in the channel

Fig. 18. A. A diagrammatic representation of the luminal surface of an upper tubule cell.The microvilli are stippled, the extracellular luminal space is white. The arrows indicatethe pathways of water movement during fluid transport. See the text for assumptions andsupporting evidence, B. A graphical representation of equation (2) (see text). The rate offluid flow increases linearly with distance in the channel, reaching a maximal value at thechannel openings. The calculated value for this maximal flow rate in the microvilli of theupper tubule is 12-5 jim/s. c. A graphical representation of fluid residence time in theintermicrovillar channels as a function of the point of entry (equation (3)). The medianresidence time is equivalent to the residence time of fluid entering at a relative positionin the channel of 0-5 (arrow). In the upper tubule, 1 have calculated this value to be 0-4s(see text).

of its significance will depend on an accurate measurement of membrane permeabilityin the microvilli and appropriate mathematical modelling of the osmotic forces invol-ved,

134 T.J.Bradley

Hill (1975) has argued on theoretical grounds that microvilli are too short to allowtotal osmotic equilibrium by means of a standing osmotic gradient, given currentlyaccepted values for the permeabilities of membranes to water. Gupta, Hall, Maddrell& Moreton (1976) measured ionic gradients along the length of the microvilli in theupper tubule of Rhodnius, using electron-probe microanalysis of frozen sections.They attempted to differentiate intra- and extracellular values, recognizing the limita-tions imposed by a probe diameter of 100 nm. It is clear from the results of the presentstudy that the intermicrovillar spaces are much too narrow (16 nm) to be resolved bythis technique. Therefore, although the study by Gupta et al. (1976) provides veryvaluable information regarding the intracellular concentrations of Na+, K+ and Cl~,it does not indicate whether the intermicrovillar spaces in the upper tubule containa standing osmotic gradient.

Maddrell (1980) has shown that the entire upper tubule epithelium has a very highosmotic permeability (>10~3 cms"1 osmol"11). It remains to be seen whether this isdue to unusual osmotic properties of the plasma membranes, or to their highly foldedconfiguration both apically and basally. Presumably, measurements of the per-meabilities of plasma membrane vesicles and more sophisticated mathematicalmodels will eventually produce answers to these questions. The morphometricanalyses of the upper tubule microvilli, presented in this paper, should be of assistancein this process.

In summary, I propose that the principal functional significance of the microvillarclumping observed in the upper tubule lies in the resultant narrow intermicrovillarspaces. In view of the current evidence that the microvilli are the site of ion transportinto the lumen, and of the slow rate of fluid flow through the channels, I propose thatthis configuration serves to maximize solute-water coupling during diuresis. Whetherthis equilibration occurs essentially instantaneously or by means of a standing osmoticgradient between the microvilli remains unanswered.

I would like to thank William Satmary and Gayle Di Carlantonio for excellent technical assis-tance. I am grateful to Dr Robert Josephson for comments on the manuscript and to Drs SimonMaddrell and Stephen Wright for valuable discussions. This work was supported by NIH grantGM-27919.

REFERENCES

BERRIDGE, M. J. & OSCHMANN, J. L. (1969). A structural basis for fluid secretion by Malpighiantubules. Tissue & Cell 1, 247-272.

BERRY, C. A. & BOULPAEP, E. L. (1975). Nonelectrolyte permeability of the paracellular pathwayin Necturus proximal tubule. Am.jf. Physiol. 22S, 581-S9S.

BOYDE, A. & WOOD, C. (1969). Preparation of animal tissues for surface scanning electron micro-scopy. J . Microscopy 90, 221-249.

BRADLEY, T. J. (1983). The excretory system: structure and physiology. In Comprehensive InsectPhysiology, Biochemistry and Pharmacology (ed. G. A. Kerkut & L. 1. Gilbert). Oxford, NewYork: Pergamon (in press).

BRADLEY, T. J. & SATIR, P. (1979). Insect axopods. J. Cell Sci. 35, 165-175.BRADLEY, T. J. & SATIR, P. (1981). 5-Hydroxytryptamine-stimulated mitochondrial movement

and microvillar growth in the lower Malpighian tubule of the insect, Rhodnius prolixus. J. CellSci. 49, 139-161.

Microvillar design in Rhodnius 135

DIAMOND, J. M. (1979). Osmotic water flow in leaky epithelia. J. Membrane Biol. 51, 195-216.DIAMOND, J. M. & BOSSERT, W. H. (1968). Standing gradient osmotic flow. A mechanism for

coupling of water and solute-transport in epithelia. J. gen. Physiol. 50, 2061-2083.FISCHBARG, J., WARSHAVSKY, C. R. & LIM, J. J. (1977). Pathways for hydraulically and

osmotically-induced water flows across epithelia. Nature, Land. 266, 71-74.GUPTA, B. L., HALL, T. A., MADDRELL, S. H. P. & MORETON, R. B. (1976). Distribution of ions

in a fluid-transporting epithelium determined by electron probe X-ray microanalysis. Nature,Lond. 264, 284-287.

GUPTA, B. J., WALL, B. J., OSCHMAN, J. L. & HALL, T. A. (1980). Direct microprobe evidenceof local concentration gradients and recycling of electrolytes during fluid absorption in the rectalpapillae of Calliphora. J. exp. Biol. 88, 21-47.

HARVEY, W. R. (1980). Water and ions in the gut. In Insect Biology in the Future (ed. M. Locke& D. S. Smith), pp. 105-124. New York: Academic Press.

HILL, A. E. (1975). Solute-solvent coupling in epithelia. A critical examination of the standing-gradient osmotic flow theory. Proc. R. Soc. Lond. B 190, 99-114.

HILL, A. E. (1977). Fluid transport across Necturus gallbladder epithelium. In ComparativePhysiology: Water, Ions and Fluid Mechanics (ed. K. Schmidt-Nielsen, L. Bolis & S. H. P.Maddrell), pp. 183-214. Cambridge University Press.

KROGH, A. (1929). Progress of physiology. Am.J. Physiol. 90, 243-251.MADDRELL, S. H. P. (1964). Excretion in the blood-sucking bug, Rhodniusprolixus Stal. II. The

normal course of diuresis and the effect of temperature. J. exp. Biol. 41, 163—176.MADDRELL, S. H. P. (1969). Secretion by the Malpighian tubules of Rhodnius. The movements

of ions and water, jf. exp. Biol. 51, 71-97.MADDRELL, S. H. P. (1971). The mechanisms of insect excretory systems. Adv. Insect Physiol. 4,

69-211.MADDRELL, S. H. P. (1978). Physiological discontinuity in an epithelium with an apparently

uniform structure. J. exp. Biol. 75, 133-145.MADDRELL, S. H. P. (1980). Characteristics of epithelial transport in insect Malpighian tubules.

In Current Topics in Membranes and Transport (ed. F. Bonner & A. Kleinzeller), pp. 427-463.New York: Academic Press.

MADDRELL, S. H. P. & GARDINER, B. O. C. (1974). The passive permeability of insect Malpighiantubules to organic solutes. .7. exp. Biol. 60, 641-652.

MADDRELL, S. H. P. & PHILLIPS, J. E. (1975). Secretion of hypo-osmotic fluid by the Malpighiantubules of Rhodnius prolixus. j . exp. Biol. 62, 671-683.

PHILLIPS, J. E. (1982). Comparative physiology of insect renal function. Am. J. Phvsiol. 241,R241-R257.

SATIR, P. (1963). Studies on cilia. The fixation of the metachronal wave. J. Cell Biol. 18, 345—365.SPRING, K. R. & HOPE, A. (1978). Size and shape of the lateral intercellular spaces in a living

epithelium. Science, NY. 200, 54-57.WALL, B. J. (1971). Local osmotic gradients in the rectal pads of an insect. Fedn proc. FednAm.

Socs exp. Biol. 30, 42-48.WHITTEMBURY, G., VERDE DE MARTINEZ, C , LINARES, H. & PAZ-ALIAGA, A. (1981). Solvent

drag of extracellular solute in leaky epithelia. In Epithelial Ion and Water Transport (ed. A. D.C. Macknight & J. P. Leader), pp. 155-164. New York: Raven.

WIGGLESWORTH, V. B. (1931). The physiology of excretion in a blood-sucking insect, Rhodniusprolixus (Hemiptera, Reduviidae). II. Anatomy and histology of the excretory system. J. exp.Biol. 8, 428-442.

WIGGLESWORTH, V. B. & SALPETER, M. M. (1962). Histology of the Malpighian tubules in Rhod-nius prolixus Stal (Hemiptera). J . Insect Physiol. 8, 299-307.

ZWICKY, K. & WIGGLESWORTH, V. B. (1956). The course of oxygen consumption during themoulting cycle of Rhodnius prolixus Stal (Hemiptera). Proc. R. ent. Soc. Lond., ser. A. 31,153-160.

(Received 25 March 1982)