1. Tubule fluid:medium ratios (TF/M) have been measured for inulin, glucose, LMWD and HMWD. These TF/M ratios were surprisingly high.

  2. The tubule appears to act as a molecular filter; that is to say, molecules move through the tubule wall in inverse relation to their size. This is best illustrated using polyvinyl pyrrolidone as a tracer. The molecular size distribution of PVP fractions present in tubule fluid differs markedly from the molecular size distribution of PVP in the bathing Ringer.

  3. No correlation can be made between the inulin and glucose TF/M and the rate of fluid production. However, the inverse relationship between TF/M and rate of fluid production for dextrans indicates a molecular sieving effect.

  4. The significance of these results is discussed with reference to models of fluid transport.

Past studies of the permeability characteristics of Malpighian tubules have been extremely limited. This is, perhaps, rather surprising, as knowledge of the permeability of a fluid-transporting tissue is of considerable value when attempting to account for the pathway of fluid movement through that tissue. The work of Ramsay (1958) showed that the tubules of the stock insect, Carausius morosus, were more permeable to amino acids and sugars than was expected. These results led to the postulation of the so-called ‘secretion/diffusion’ theory of Malpighian tubule excretion. Ramsay concluded that ‘the tubule is a means whereby all soluble substances of low molecular weight are indiscriminately removed from the haemolymph’. However, there have been no further studies to identify the true nature of such ‘low molecular weight’ substances and the restrictive properties of the Malpighian tubule epithelium. The results of the present study have shown that to assume that the tubule is only permeable to substances of ‘low molecular weight’ is erroneous, and that a fresh look at the mechanism of fluid production is required in this case.

Tubules dissected from specimens of Glomeris were isolated in serum or Ringer as described previously (Farquharson, 1973 a). Radioactively-labelled compounds of varying mol. wt were added in measured aliquots (500 nl) to the medium bathing the isolated tubules. After an appropriate period the radio activities of measured aliquots of tubule fluid and medium were estimated. Radioactivity was measured with either a Tracerlab automatic gas-flow Geiger counter or with a Nuclear Enterprises liquid scintillation counter type 6500A. The compounds used were:

[3H]lnulin and [14C]inulin (mol. wt ca. 5000).

[14C]Dextran of ‘Low’ mol. wt (16000–19000); hereafter designated as LMWD.

[14C]Dextran of ‘high’ mol. wt (60000–90000); hereafter designated as HMWD.

[125I] Polyvinyl pyrrolidone (PVP) (mol. wt 5000–80000; average 40000).

[14C]D-Glucose and [14C]-L-glucose (mol. wt 180).

The dextrans were bought from New England Nuclear Corp., Frankfurt, Germany. All other labelled compounds were obtained from the Radiochemical Centre, Amersham.

Gel-filtration chromatography was used to determine the molecular size distribution of [125I]-PVP before and after passing through the isolated tubules of Glomeris. This was carried out using Sephadex G200 (Pharmacia U.K. Ltd) in small columns (16·5 cm long × 0·5 cm diameter) similar to those described by Riegel (1966). The columns were eluted with 0·5 % NaCl + 0·05 % sodium azide. A 2 μl sample of tubule fluid or Ringer was placed on the surface of the gel bed. Flow was begun and the first 30 drops were collected individually on planchettes. The planchettes were air-dried, and their radioactivity was measured in a gas-flow Geiger counter.

Tubule fluid and Ringer were subjected to thin-layer chromatography to determine whether or not there was any degradation of labelled compounds. Labelled glucose was treated as follows: 2μl samples were spotted on aluminium foil plates that had been pre-coated with Kieselguhr (250 μm thick, Merck). The plates were placed in glass TLC tanks which had equilibrated with a solvent composed of 65 ml ethyl acetate:35 ml isopropanol/water mixture (2:1) for 24 h. The dried chromatogram was apposed to X-ray film between two glass plates for 10 days, after which the presence of labelled compounds was visualized by developing the X-ray film. A comparison with standard solutions revealed the extent to which D-and L-glucose were subject to metabolization.

Gel-column chromatography was utilized to determine whether or not there was any decomposition of inulin. 1 μl samples of Ringer or tubule fluid containing [14C]inulin were passed through a small column packed with Sephadex G 25 and eluted with 0·5 % NaCl solution. The first 30 fractions were collected individually on planchettes and the radioactivity was measured. The results of these experiments were compared with the results of similar experiments carried out on standard inulin solutions.

(1) Permeability of the tubules

Initially, the permeation of the tubule by the polyfructoside inulin (M.W. 5000) was measured. This test substance was chosen because of its value in the study of vertebrate and invertebrate excretory systems as a compound which is thought to be neither secreted nor reabsorbed (Riegel, 1972). Previous investigations by Ramsay & Riegel (1961) using the tubules of Carausius implied that the permeability of Malpighian tubules to molecules of this size was very low (TF/M = 0·046). However, the TF/M values obtained in the present study were in some cases as high as 0·97 and averaged 0·68 (see Table 1). Comparison of the elution pattern of [14C]inulin in samples of Ringer and tubule fluid subjected to gel filtration revealed that there was no detectable alteration of inulin whilst passing through the tubule.

The surprisingly high TF/M for inulin (Table 1) indicated that similar studies involving larger mol. wt compounds would be necessary in order to obtain a better picture of the limiting properties of Glomeris tubules. Dextrans (polyglucosides) were employed in a manner similar to that used for inulin to characterize the tubule permeability. Dextrans were first used by Brewer (1951), who demonstrated the inverse relationship between molecular size and the rate of movement of dextran across the glomerular membranes of rabbits. Dextrans have also been utilized extensively by Wallenius (1954) to study glomerular permeability in dogs. Average TF/M ratios for dextrans obtained in the present study are shown in Table 1. Tubules of Glomeris severely limit the passage of HMWD, but there is much less restriction to the movement of LMWD.

The foregoing results, taken together with the high permeability of the tubules to glucose, suggested that carbohydrate molecules move through the tubule wall in inverse relation to their size. The inverse relationship between permeability and mol. wt was more readily illustrated by using polyvinyl pyrrolidone as a tracer molecule; polyvinyl pyrrolidone is a substance of uniform chemical composition which exists as a range of polymers. In the present case, the PVP used had an average mol. wt of 40000, but the mol. wt range was 5000–80000. Hardwicke et al. (1968) injected trace amounts of the polymer into rabbits and then determined the molecular size distribution in plasma and urine by gel filtration. From their results they were able to get an approximation of the glomerular permeability. In the present work 2 μl of a concentrated solution of PVP were added to 15 μl of Ringer bathing the tubules. Gel filtration was carried out on three samples each of Ringer and tubule fluid. The tubule fluid samples represent the volume of fluid pooled from 15 tubules. The elution patterns of PVP in Ringer and tubule fluid were compared with the elution patterns of inulin and HMWD. This comparison provided a rough guide to the mol. wt of PVP to be expected in each elution sample. The results of these studies are presented in Text-fig. 1.

As shown in Text-fig. 1 the molecular size distribution of PVP fractions present in tubule fluid differs markedly from the molecular size distribution of PVP in the bathing Ringer. Thus, the peak in radioactivity that was observed in the early fractions (80000 M.w. range) when Ringer was chromatogramed, shifted to the right when tubule fluid was chromatogramed; the radioactivity reached a maximum in fractions corresponding to a mol. wt of 5000. Therefore, there was a progressive increase in TF/M with decreasing molecular size of PVP fraction, as was true with the carbohydrates. Also of interest is the fact that the Malpighian tubule of Glomeris is permeable to the complete mol. wt range (5000–80000). However, as indicated in the dextran studies, the permeability to fractions of the highest mol. wt is very low. There appears to be no sharp cut-off point with increasing molecular size.

TF/M ratios of unity observed for D-and L-glucose are consistent with the inulin and dextran studies ; that is to say, the restriction on the permeability of the tubule to a molecule of this size is negligible. There was a depletion in the amount of labelled D-glucose in the bathing Ringer compared with the values for L-glucose, suggesting that D-glucose was metabolized. The results of chromatogramming both isomers in Ringer and tubule fluid have shown that this is, in fact, the case. Two extra spots were visible after chromatogramming D-glucose in the Ringer and tubule fluid. None of the L-glucose would appear to have been metabolized.

(2) The effect of fluid production rate on permeability

The progressive deterioration in secretion rate normally observed in tubules of Glomeris (Farquharson, 1973 a) facilitated the estimation of TF/M ratios for various polymers over a wide range of secretion rates using the same tubule. However, the results of such experiments (Text-fig. 2) have shown that no correlation could be made between the inulin TF/M ratio and the rate of fluid production, even when the two parameters were measured on the same tubule. TF/M ratios for D-and L-isomers of glucose (Text-fig. 3) also bore no obvious relationship to the rate of fluid production.

Text-figs. 4, 5 reveal that there was an inverse relationship between TF/M and rate of fluid production for dextrans. This effect was found to be accentuated when the TF/M for HMWD at varying fluid production rates was measured. HMWD can be excreted with a TF/M of up to 0·2 at low rates of fluid secretion.

Patton & Craig (1939) found that the Malpighian tubules of Tenebrio molitor are readily permeable to glycine, glutamic acid, uric acid, and urea. More recently, Ramsay (1958) has shown that the TF/M of exogenous amino acids, sugars and urea across isolated tubules of Carausius are never greater than one. Furthermore, the TF/M for these substances was independent of their concentration in the bathing medium. Ramsay suggested that these results, plus the fact that there was no interference between different substances, were best explained by a passive diffusion process.

The Malpighian tubules of Glomeris appear to be more permeable to carbohydrates than are insect tubules. For example, the inulin TF/M for tubules of Carausius averaged only 0·046 (Ramsay & Riegel, 1961). In insects other than Carausius the average TF/M may be higher (e.g. Calliphora, 0·46 ; M. R. Phillips personal communication, Musca domestica, 0·30; D. M. Windmill personal communication). Nevertheless, the average inulin TF/M of o·68 seen in Glomeris is very high considering that it was measured during relatively rapid fluid production. The high permeability of tubules of Glomeris was perhaps best illustrated by the experiments using PVP in which estimations of TF/M over a wide continuous spectrum of molecular sizes were made. Laurent & Killander (1964) and Hardwicke et al. (1968) have shown that for PVP there is an inverse linear relationship between the fraction of the compound in the gel phase of the chromatography column and the logarithm of the molecular radius over the size range of 15–80 Å. Therefore, the peak of radio-activity observed in late fractions when tubule fluid was placed in Sephadex gas Columns was a direct expression of an increased permeability of the tubules to molecules with a smaller radius. Malpighian tubules would appear to act as molecular filters.

Selection of smaller molecules by tubules of Glomeris is unlikely to be due solely to a passive diffusion process. The permeability of the tubules to various nonelectrolytes within a wide range of mol. wts (Table 1, Text-figs. 4, 5) and over a range of secretion rates indicates a molecular sieving effect. The degree of molecular sieving of solute depends upon the ratio of its restricted diffusion coefficient to the rate of volume flow (filtration) through the membrane (Pappenheimer, 1953). The diminished TF/M values for dextrans at high rates of fluid flow are consistent with the hypothesis that the faster the rate of fluid movement through pores, the greater will be the likelihood that particles will strike the pore edge and be retarded. Therefore, at high rates of fluid flow, diffusion is important in determining the rate of solute movement. At low rates of fluid transport, filtration plays a relatively more important role in determining the passage of these molecules across the tubule wall, with a resultant higher TF/M ratio. It is seen, then, that molecular sieving results from an imbalance between filtration and diffusion across a membrane or epithelium. However, for large lipid-insoluble molecules (such an inulin and dextrans) the restriction to diflfusion becomes so great that the degree of molecular sieving is determined largely by the rate of filtration. In theory it is not difficult to reconcile the concept of filtration with fluid production by the in vitro Malpighian tubule. That is, it is unnecessary to assume that the required differential in hydrostatic or osmotic pressure across the tubule wall must be generated external to the epithelium. However, the present results are not readily explicable within the framework of the secretion-diffusion theory of fluid production by Malpighian tubules (Ramsay, 1958).

In the discussion of these results it is relevant to question the route of large inert molecules across the tubule wall. The permeability of the tubule to inulin and dextrans was considerably higher than expected, and it is unlikely that their trans-tubular route is through the cell. Such an explanation would necessitate enormous energy expenditure and deformation of the cell plasma membrane to allow molecules with a radius of 15 Å to reach TF/M ratios of o·68 or greater. Study of the ultrastructure of the tubule (see Plate 1, and Farquharson, 1972) suggests that the most obvious route is through the intercellular junctions, whose diameter (300 Å) is compatible with this suggestion. Furthermore, it seems likely that the septa of the intercellular apical junctions do not form a continuous connexion around the cell perimeter. There is increasing evidence that the intercellular spaces are pathways of solute and water movement in a variety of epithelia. For example, Fromter & Diamond (1972) have shown that in Necturus gallbladder 95 % of the trans-epithelial electrical conductance must be in a shunt by-passing the cells. The total resistance across the apical and basal cellular membranes was far in excess of the measured trans-epithelial resistance. From a survey of various vertebrate tissues they concluded that only ‘tight’ epithelia can maintain steep salt gradients, have high trans-epithelial electrical resistances and a large electrical potential difference between symmetrical bathing solutions. The tubules of Glomeris obviously are ‘leaky’, even with respect to the movement of nonelectrolytes ; therefore, if the analysis of Fromter and Diamond is correct, the observed lack of an electrical potential difference, the lack of solute gradients and a trans-epithelial osmolarity ratio of unity (Farquharson, 1973 b) are all to be expected in Glomeris.

It is believed generally that fluid transport across epithelia is linked to active transport of small solutes. A number of models have been proposed in order to explain the link-up of water and solutes. The theory of standing gradient osmotic flow was proposed by Diamond and his co-workers (Diamond & Bossert, 1967, 1968; Tormey & Diamond, 1967) to account for fluid re-absorption by the gallbladder, and since has been applied to so-called ‘backwards transporting’ epithelia (i.e. fluid secreting) such as the Malpighian tubules (Berridge, 1968; Berridge & Oschman, 1969; Diamond & Bossert, 1968; Maddrell, 1969, 1971). Essentially, the theory proposes that osmotic gradients are established in channels within the transporting epithelium such as basal infoldings, microvilli and intercellular spaces. Water moves passively across cellular membranes into the spaces in response to osmotic gradients. In early work on the ‘standing gradient’ model it was assumed that the intercellular junctions in ‘forwards transporting’ epithelia were ‘tight’; they appeared to be virtually impermeable to water-soluble molecules with diameters greater than ca. 6 A. However, the more recent electrical evidence (e.g. Fromter & Diamond op. cit.) does not support this.

Doubts arise when the ‘standing gradient’ model is applied to the short channels (5–10 μm) of the Malpighian tubule. Diamond & Bossert (1968) have shown that channel hypotonicity will be more marked the greater its length, the more rapid the rate of solute transport, the lower the water permeability and the smaller the channel radius. When considering the structure of the Malpighian tubules of Glomeris one is presented with an arrangement of fairly short microvilli and not only short (ca. 3–5 μm) but wide (ca. 3000 Å) extracellular channels which do not appear to end blindly (Plate 1 and Farquharson, 1972). Also, due to extensive interdigitation of the cells, the junctional area of the luminal surface is much larger than that found in other Malpighian tubules where the cells are larger. The original hypothesis was based on the compartments being unstirred to allow the formation of standing osmotic gradients. It has been suggested that unulin and dextrans probably move through the junctions rather than through the cell. The high TF/M ratios for these substances measured across the tubule of Glomeris suggest that the intercellular passage is rather more than a ‘leak’ pathway (as suggested by Dibona (1972) for the junctions of toad skin or bladder) and would probably be better described as the major fluid pathway. Considering the high rates of fluid production, there appears to be no conceivable way by which a gradient could be established within the channel. Even if one could envisage some form of gradient, then according to the basic hypothesis the fluid moving through the junctions would tend towards hypotonicity. However, the tubule fluid produced by isolated Malpighian tubules of Glomeris is isosmotic to the bathing medium although there is some tendency for the tubule fluid to be hyperosmotic to the medium. Given the permeability characteristics of the Malpighian tubules of Glomeris, it is difficult to envisage the way in which a localized osmotic gradient could be established as required by standing-gradient osmotic flow.

Other mechanisms of fluid transport which have been proposed, however, are still based on solute being pumped into a confined space (Taylor, 1971 ; Curran, 1960, 1965). Taylor ‘s model takes no account of the possibility of fluid movement through tracellular channels (a ‘leaky’ system provides no restriction to back diffusion of solute from the lumen). The Curran model suffers from the same disadvantage as other models: solute pumped into the intercellular space would rapidly diffuse away. Therefore, in order to envisage such systems in operation it would also be necessary to postulate a very fast rate of solute pumping. However, the latter model could be utilized to explain the molecular sieving of high mol. wt compounds seen in the present work. A similar model of trans-epithelial fluid movement has been suggested by Riegel (1970b). This ‘formed body’ model is based on the swelling of formed material (possibly lysosomes) within a restricted space and creating an increase in osmotic pressure there. (Riegel (1970 a) has shown that when formed bodies swell, water, but not solutes, moves into them.) Water will enter the space accompanied by solute, giving rise to an increase in hydrostatic pressure. Thus, fluid will move out of the space in a direction which is determined by the restriction to fluid movement of the access passages. This theory is similar to those of both Curran and Diamond and their associates. However, the main difference, the manner in which solute is localized within spaces, is of prime importance to views of fluid transport across very ‘leaky’ epithelia such as the Malpighian tubules of Glomeris. No formed bodies have been seen in the extracellular space (using electron microscopy) except in ‘non-functional’ tubules. It may be argued that this is simply a result of the formed bodies bursting at fast rates of fluid production, so that formed bodies are visible only in tissue that is secreting at extremely low rates. However, this model can conveniently explain the results obtained with high mol. wt compounds and the production of an isomolar fluid (Farquharson, 1973b).

The foregoing discussion has but presented and considered the alternative possibilities with respect to the mechanism underlying fluid production by isolated Malpighian tubules of Glomeris. Further investigations are required. However, as mentioned previously (Farquharson, 1973 b), the structural and permeability characteristics of this tubule should be taken into account when proposing the involvement of solute pumps in such a mechanism. Many features of the physiology of the tubules of Glomeris are unique amongst Malpighian tubules and therefore they do not fit easily into previous conceptions of tubule function.

This paper represents part of a thesis submitted to the University of London for the degree of Ph.D. I would like to express my sincere gratitude to Dr J. A. Riegel for his constructive discussion and supervision of the work and for his helpful criticism of the manuscript. I am also very grateful for the assistance that my colleagues Drs M. A. Cook, T. Dalton, and J. A. Riegel and Mr M. R. Phillips have given in the collection of specimens, and I would like to thank Mr M. R. Phillips and Mr D. M. Windmill for allowing me to quote from their unpublished studies. I am indebted to the Medical Research Council for their financial support of this work.

Berridge
,
M. J.
(
1968
).
Urine formation by the Malpighian tubules of Calliphora. I. Cations
.
J. exp. Bid
.
48
,
159
74
.
Berridge
,
M. J.
&
Oschman
,
J. L.
(
1969
).
A structural basis for fluid secretion by Malpighian tubules
.
Tissue and Cell
1
,
247
72
.
Brewer
,
D. B.
(
1951
).
Renal clearances of dextrans of varying molecular weights
.
Proc. R. Soc. Med
.
44
,
561
3
.
Curran
,
P. F.
(
1960
).
Na, Cl and water transport by rat ileum in vitro
.
J. gen. Physiol
.
43
,
1 137
48
.
Curran
,
P. F.
(
1965
).
Ion transport in intestine and its coupling to other transport processes
.
Fedn Proc. Fedn Am. Socs exp. Biol
.
24
,
993
9
.
Diamond
,
J. M.
&
Bossert
,
W. H.
(
1967
).
Standing gradient osmotic flow. A mechanism for coupling of water and solute transport in epithelia
.
J. gen. Physiol
.
50
,
2061
83
.
Diamond
,
J. M.
&
Bossert
,
W. H.
(
1968
).
Functional consequences of ultrastructural geometry in ‘backwards’ fluid-transporting epithelia
.
J. Cell Biol
.
37
,
694
702
.
Dibona
,
D. R.
(
1972
).
Passive intercellular pathway in amphibian epithelia
.
Nature new Biol
.
238
,
179
81
.
Farquharson
,
P. A.
(
1972
).
The function and microstructure of the Malpighian tubules in Glomeris marginata (V
.).
Ph.D. Thesis
,
University of London
.
Farquharson
,
P. A.
(
1973a
).
A study of the Malpighian tubules of the pill millipede, Glomeris marginata (Villers). I. The isolation of the tubules in a Ringer solution
.
J. exp. Biol
.
60
,
13
28
.
Farquharson
,
P. A.
(
1973b
).
A study of the Malpighian tubules of the pill millipede, Glomeris marginata (Villers). II. The effect of Na, K and osmotic pressure on fluid production
.
J. exp. Biol
.
60
,
29
39
.
Frömtbr
,
E.
&
Diamond
,
J. M.
(
1972
).
Route of passive ion permeation in epithelia
.
Nature new Biol
.
235
.
9
13
Hardwicke
,
J.
,
Hulme
,
B.
,
Jones
,
J. H.
&
Ricketts
,
C. R.
(
1968
).
Measurement of glomerular permeability to polydisperse radioactively labelled macromolecules in normal rabbits
.
Clin Sci
.
34
,
505
14
.
Laurent
,
J. C.
&
Killander
,
J.
(
1964
).
A theory of gel filtration and its experimental verification
.
J. Chromatogr
.
14
,
317
30
.
Maddrell
,
S. H. P.
(
1969
).
Secretion by the Malpighian tubules of Rhodnius. The movements of ions and water
.
J. exp. Biol
.
51
,
71
97
.
Maddrell
,
S. H. P.
(
1971
).
The mechanisms of insect excretory systems
.
Adv. Insect Physiol
.
8
,
199
331
Pappenheimbr
,
J. R.
(
1953
).
Passage of molecules through capillary walls
.
Physiol. Rev
.
33
,
387
423
.
Patton
,
R.
&
Craig
,
R.
(
1939
).
The rates of excretion of certain substances by the larvae of the mealworm, Tenebrio molitor (L)
.
J. exp. Zool
.
81
,
437
57
.
Ramsay
,
J. A.
(
1958
).
Excretion by the Malpighian tubules of the stick insect, Dixippus morosus (Orthoptera, Phasmidae): amino acids, sugars and urea
.
J. exp. Biol
.
35
,
871
91
.
Ramsay
,
J. A.
&
Riegel
,
J. A.
(
1961
).
Inulin excretion by Malpighian tubules
.
Nature, Lond
.
191
,
1115
.
Riegel
,
J. A.
(
1966
).
Analysis of formed bodies in urine removed from the crayfish antennal gland by micropuncture
.
J. exp. Biol
.
44
,
387
95
.
Riegel
,
J. A.
(
1970a
).
In vitro studies of fluid and ion movements due to swelling of formed bodies
.
Comp. Biochem. Physiol
.
35
,
843
56
.
Riegel
,
J. A.
(
1970b
).
A new model of transepithelial fluid movement with detailed application to fluid movement in Crayfish antennal gland
.
Comp. Biochem. Physiol
.
36
,
403
10
.
Riegel
,
J. A.
(
1972
).
Comparative physiology of renal excretion
.
In University Reviews in Biology
. Editor,
J. E.
Treherne
.
London
:
Oliver & Boyd
.
Taylor
,
H. H.
(
1971
).
Water and solute transport by the stick insect, Carausius morosus. The normal ultrastructure of type I cells
.
Z. Zellforsch. mikrosk. Anat
.
118
,
333
68
.
Tormey
,
J. M.
&
Diamond
,
J. M.
(
1967
).
The ultrastructural route of fluid transport in rabbit gall bladder
.
J. gen. Physiol
.
50
,
2031
60
.
Wallenius
,
G.
(
1954
).
Renal clearance of dextran as a measure of glomerular permeability
.
Acta Son Med. upsal. Suppl
.
4
.

PLATE I

(For preparation of electron micrographs see Farquharson, 1972).

(a) Cross-section through the wall of the upper tubule showing the numerous septate junctions (sj) and the large intercellular spaces (ics). × 21000.

(b) Tangential section through the septate junctions. Note the lack of septa in one of the junctions, ×70000.

(c) Permeation of lanthanum (M.W. 433) into the junctions, × 104000.

bm, basement membrane; GA, Golgi apparatus; er., endoplasmic reticulum; ics, intercellular space; L, lumen; la, lanthanum; m, mitochondria; mlb, multilaminate body; mv, microvilli; n, nucleus; sj, septate junction.