Clinical Science andMolecular Medicine (1978) 55, 113-1 19
Effects of concentration-polarization on the filtration of
proteins through filters constructed from isolated renal
basement membrane
T. G. COTTERAND G . B. R O B I N S O N
Department of Biochemistry, University of Oxford, U.K.
(Received 29 September 1977; accepted 9 March 1978)
Summary
1. Filters comprising multiple layers of rabbit
renal tubular basement membrane were constructed with conventional pressure filtration chambers.
The effects of concentration-polarization on the
behaviour of these filters was assessed by studying
the filtration of proteins and of serum under
turbulent (stirred) and unstirred conditions.
2. With stirring bovine serum albumin was
effectively rejected by the filter barriers (a = 0.95)
but rejection was diminished (a = 0.18) without
stirring. The hydraulic permeability of the filters
also fell without stirring.
3. In the presence of horse immunoglobulin G, a
wholly rejected protein, the rejection of cytochrome c was increased and hydraulic flux was
reduced.
4. Filtration studies of serum showed that serum
protein was effectively rejected with stirring (a >
0.999) but rejection diminished when stirring
ceased (a = 0.98). Albumin was the only protein
detected in the filtrate with stirring but a- and bglobulins appeared when stirring ceased.
5. These results show that concentrationpolarization markedly affects the behaviour of
these basement membrane filters in uitro, since
without stirring a polarization layer of rejected
protein is formed, which reduces hydraulic permeability and results in increased protein permeation through the filter.
Key words: basement membrane, concentrationpolarization, glomerular filtration, 1 kidney,
proteinuria.
Introduction
The wide use of filters in industrial processes has
focussed attention on the manner in which their
behaviour in liquid-phase processes can be affected
by concentration-polarization (Blatt, Dravid,
Michaels & Nelsen, 1970; Goldsmith, 1971;
Dejmek, 1975). Concentration-polarization results
from the accumulation of rejected solute molecules
at the filter face and this unstirred layer modifies
the properties of filters. In the simplest case, an
accumulated layer of wholly rejected macromolecules at the filter surface reduces the flux of
solvent through the filter; in essence a second filter
barrier is created on the surface of the filter
membrane so that overall behaviour is determined
by the dual barrier. For partially rejected solutes
the effects of concentration-polarization are more
complex, for the polarization layer results in a concentration of solute molecules at the entries to the
channels that accept them, and thus the concentration of solute in the filtrate increases. The
formation of the polarization layer depends upon
the rate at which solute is carried to the filter
surface by the filtration flux, and the rate at which
the layer tends to disperse as a result of back
diffusion assisted by stirring (Blatt et al., 1970).
The glomerulus functions as an ultrafilter of
blood. Mathematical models have been devised to
explain the permeability properties of the
Corresoondence: Dr G. B. Robinson. Deoartment of Biochemistry, University of Oxford, South'Parlk Road, Oxford
OX1 3QU, U.K.
8
113
114
T. G. Cotter and G. B . Robinson
glomerular barrier (Renkin & Gilmore, 1973) and
recent models, based on hydrodynamic or thermodynamic theory, have taken account of the fact
that glomerular filtration occurs progressively
along a tubular capillary (Chang, Robertson, Deen
& Brenner, 1975; Dubois, Decoodt, Gassee,
Verniory & Lambert, 1975). However, none of
these general theories has made allowance for the
effects of concentration-polarization, although
there is one theoretical appraisal (Deen, Robertson
& Brenner, 1974). Morphological studies demonstrating that the passage of albumin across the
glomerular capillary wall was influenced by vascular blood flow led to the suggestion that concentration-polarization may influence the permeability characteristics of the glomerulus (Ryan
& Karnovsky, 1976).
Here we demonstrate the effects of concentration-polarization on the filtration behaviour
of ultrafilters prepared from renal basement membrane (Robinson & Brown, 1977). Although this
experimental model is unphysiological, it permits
filtration conditions to be more closely controlled
than is possible in vivo. Thus the effects of concentration-polarization on the filtration of macromolecular solutes through basement membranes
can be tested empirically by changing the stirring
conditions during filtration.
Materials and methods
Basement membranes were isolated from freshly
dissected rabbit kidney cortex by a procedure
which has been described in detail elsewhere (Ligler
& Robinson, 1977). The preparations, which consisted of a mixture of tubular and glomerular
membranes, were contaminated to a minor extent
by collagen fibres but were devoid of cellular
elements. The basement membrane filters were constructed by packing the membrane fragments into
coherent layers on a filter support in Amicon
pressure filtration chambers of either 10 ml or 65
ml capacity (Amicon Ltd, High Wycombe, Bucks.,
U.K.), which were assembled with a cellulose
acetate membrane [Millipore (U.K.) Ltd, London;
0.47 pm exclusion] lying on a hardened Whatman
no. 50 filter paper, which in turn rested on the
porous polypropylene base support. After assembly the filtration chambers were filled with buffered
saline [sodium chloride (0.15 mol/l)/Tris/hydrochloric acid buffer (0.01 mol/l), pH 7-41 and a
portion of the basement membrane suspension was
added; approximately 1.5 mg of protein was used
for the large chamber (filtration area 13.8 cm’) and
approximately 0.5 mg in the smaller chamber
(filtration area 4.9 cm’). Pressure was applied to
the chambers (with N, gas at 200 kPa). As the
membrane suspension passed through the filter, the
particles built up on the Millipore membrane
surface to form a thin compact layer, which progressively slowed the solvent flow. Filtration was
continued until all the liquid had passed through
the filter and at this point more buffered saline was
added to the chamber and filtration was continued
for a further 30 min. Filtration of the proteins and
of serum was subsequently conducted at 150 kPa
pressure, filtrate being collected over timed intervals. All experiments were conducted at 2OOC.
During filtration the chamber contents were stirred
magnetically at 1000 rev./min, when turbulent
stirring was achieved (Blatt et al., 1970). All the
solutions used for filtration, other than the basement membrane suspension, were first filtered
through a Millipore membrane (0.47 p m exclusion).
For filtration experiments, either purified proteins dissolved in buffered saline, or rabbit serum,
were used; pure proteins were obtained commercially (Sigma London Chemical Co. Ltd, Kingstonupon-Thames, Surrey, U.K.) and serum from
rabbit heart blood was taken after death. Concentrations of the purified proteins were measured
by absorbance (A2,,,,) of the solutions. For serum,
the total protein concentration was measured
(Lowry, Rosebrough, Farr & Randall, 1951) after
precipitation of the protein with 5% (w/v) trichloroacetic acid; the precipitates were dissolved in
reagent C (Lowry et al., 1951) before colour
development. In some experiments the proteins in
filtrates and in the overstanding solutions were
examined by electrophoresis on polyacrylamide
gels (Ligler & Robinson, 1977).
In calculating the results, the porosity of the
membranes towards macromolecules is expressed
as the rejection coefficient ((I), i.e. the fraction of
solute that does not pass through the membrane:
(I = C, - C,/C,, where C,, is the concentration of
the solute in the bulk solution and C, is the concentration in the filtrate. Values are given as mean
SD.
Results
Initial experiments assessed the extent to which
macromolecules escaped through leaks in the filter
pads of basement membrane fragments, by determining the rejection coefficients of bovine serum
albumin (0.5 mg/ml) and horse immunoglobulin G
(IgG, 0.5 mg/ml) while stirring. The values of (I
115
Filtration through basement membranes
h
0°+
.E:
E
c
3
2
1
0
2
3
4
5
6
Volume filtered (ml)
7
60
8
10-
I
I
0
1
2
0
1
2
I
I
T
+ s I
3
4
5
6
Conc. of IgG (mg/ml)
7
8
6 8 c
';06-
.*
0
'30.4 -
d
0.2
I
0
1
I
2
I
I
I
3
4
5
Volume filtered (ml)
I
I
I
6
7
8
FIG.1. Effect of stirring on the filtration of serum albumin
through basement membrane filkr pads. The filter pads were
formed in small Amicon cells (capacity ID ml) and serum
albumin (initial concentration 0.5 mg/ml) was filtered across the
membrane with (0) and without (A) stirrink. Flow rates and
rejections were measured duriiig the filtration; initial flow rates
were measured with buffer alone. The rekults are the mean
yalues of nine separate filters ahd the bars indicate & SD.
obtained yere 0.95 k 0.01 and 0.985 f 0.01
respectively for six sepafate filters. It is impossible
to distinguish between seepage and porosity but as
IgG was almost entirely rejected by the filters
seepage probably did hot account for more than
1.5% of the solvent flow through the membrane.
The lower rejection fdr serum albumin indicated
that the membrane pads were porous to the smaller
protein. These results were similar to those obtained in experiments with filtration by vacuum
(Robinson & Brown, 1977).
To study concentration-polarization effects on
the filtration of proteins, filtration behaviour was
compared with and without turbulent stirring. The
rejection of serum albumin remained constant (a =
0-95) with stirring and the flow rate barely
changed; with no stirring the rejection fell (u =
0.18) and the flow rate declined (Fig. 1).
The effects of stirring on the filtration of cytochrome c in the presence of IgG were examined to
establish whether the presence of a wholly rejected
protein (IgG) affected the filtration of a partially
rejected protein (cytochrome c). Without IgG cytochrome c was rejected more effectively with stirring
(Fig. 2), as was albumin. With increasing concentrations of IgG the rejection of cytochrome c
I
I
I
I
3
4
5
Conc. of IgG (mg/ml)
I
I
I
6
7
8
FIG.2. Effect of IgG on the rejection of cytochrome c during
filtratiop through basement membrane filter pads with (0)and
without (A) stirring. The filter pads were formed in small
Amicod celis (10 ml bapacity) and the solutions filtered
contained cyiochrome c (initial concentration 0.5 mg/ml) and
IgG at inltial concentratibihs shown on the abscissae. Flow rates
and cytoEhrome c rejections were measured in duplicate with
three separate filters, and the results shown are the mean values;
the bars ihdibate f s ~ .In these experiments cytochrome c concentrations were determined from measurements of absorbance
at 550 nm.
increased, and did so more markedly with stirring.
The filtration rate fell with increasing IgG concentratiofi but this effect was more marked with no
stirring. Thus a wholly rejected protein markedly
increased the rejection of the smaller cytochrome c
but the size of this effect depended on stirring.
The effect of stirring during the filtration of
rabbit serum showed that much less protein passed
through the filters during stirring than when stirring
ceased (Fig. 3). This change depended on stirring,
as when stirring was recommenced the protein concentration of the filtrate then fell. The rejections for
whole serum protein rose during the initial stage of
the filtration with stirring to very high steady-state
values (a > 0.999). At the same time filtration
rate fell, indicating that build-up of a polarization
layer was influencing both the flow rate and
rejection. The protein 'leakage' during non-stirring
caused the protein concentration in the filtrate to be
> 1.0 mg/ml, whereas during stirring this was < 10
,ug/ml.
116
80
T. G. Cotter and G. B . Robinson
a
50
Stirring
ceased
70
i
Started
1
1
0
30
Sample no.
I
I
I
I
50
100
500
1000
500
1000
Stirrer speed ( r e v h i n )
1
1
1.000
1.oo
h
0.995
30
50
100
Stirrer speed ( r e v h i n )
0.98
I
I
I
I
I
I
I
I
I
,
1
2
3
4
5
6
7
8
9
10
Sample no.
FIG.3. Effect of stirring on the filtration of serum through basement membrane pads. The filter pads were formed in large
Amicon cells (capacity 65 ml) and serum was filtered initially
with stirring; stirring was stopped at the point indicated (arrow)
and then restarted. The lag in response of the rejection to the
cessation of stirring results from the washout of previous filtrate
from the apparatus. Initial flow rates were measured with buffer
alone. The results shown represent one experiment only, since
the different time bases used in different experiments make it
impossible to aggregate the results. Six experiments, each with
three filters, showed essentially identical results.
The proteins in the serum filtrates were analysed
by electrophoresis on polyacrylamide gels in
Tridglycine buffer, pH 8.9, without sodium
dodecyl sulphate. In filtrates collected during
stirring the predominant band was albumin,
although some very faint bands were seen close to
the origin. Filtrates collected after stirring had
FIG.4. Effect of reducing stirrer speed on the filtration of serum
through basement membrane pads; stirrer speeds were
measured electronically. The results are the mean values for
duplicate determinations obtained with three separate filters and
the bars indicate ~ S D .
ceased showed a prominent albumin band together
with other proteins which ran in the positions of aand ,&globulins, but no proteins were observed in
the y-globulin region. Thus non-stirring caused the
filters to be less selective than stirring but, even so,
no significant amounts of y-globulin escaped.
As the above changes from turbulent stirring to
no stirring were abrupt, we studied protein rejection
as stirring speed was progressively reduced, greater
quantities of protein then escaping through the filter
(Fig. 4). Thus the build-up of the polarizing layer
was not an all-or-none effect, but occurred by
gradual progression as stirring efficiency at the
filter face diminished.
Filtration through basement membranes
Discussion
Our studies provide striking demonstrations of the
effects of concentration-polarization on the
behaviour of basement membrane filters. With
albumin rejection fell to very low values in the
absence of stirring, with only a slight fall in solvent
flux through the membranes; at the same time the
albumin flux increased from 1.2 yg min-l cm-* to
16.8 yg min-’ cm-* in the absence of stirring (Fig.
1). According to concentration-polarization theory
(Blatt et al., 1970) this results from build-up of a
more concentrated-polarization layer of rejected
albumin molecules at the surface of the filter in the
absence of stirring, which reduces solvent flux.
Rejection fell, as the actual rejection values for the
filter should be expressed as u = C, - CJC,,
where C, is the albumin concentration at the filter
face. However, we measured C, not C,; hence
rejection fell as C, increased with respect to C,
during the experiment.
When cytochrome c was filtered with increasing
concentrations of IgG, this always reduced the
solvent flux, particularly in the unstirred chambers,
and cytochrome c rejection increased. This indicates that a polarization layer of IgG was formed
even during turbulent stirring, so reducing the
permeability of the filters to solvent and solute.
Without stirring the very low hydraulic permeabilities of the filters indicated the formation of a
thicker, or more concentrated, polarization layer of
IgG, but the effect of this layer on the rejection of
cytochrome c was less marked owing to the
increased wall concentration (C,) of this protein.
The changes in filter behaviour resulting from concentration-polarization thus not only depend on
stirring but also on the admixture of solutes at
differing concentrations.
When whole serum was filtered cessation of
stirring reduced solvent flux and increased the
protein concentration of the filtrate (Fig. 3). As
with albumin, protein flux through the filter
increased. This change was reversed by recommencing stirring, indicating that it arose from concentration-polarization; these effects were progressive as the stirrer speed was reduced (Fig. 4).
With stirring, albumin was the only protein that
escaped significantly through the filter, whereas
without stirring a- and /I-globulins also passed
through. This presumably resulted from the
increased wall concentration of these proteins in
the non-stirred state, although the permeability
properties of the polarization layer may differ
under different stirring conditions. When serum
117
was filtered with stirring, very little albumin (a >
0-99), and no IgG, escaped through the filters, but
when these proteins were filtered alone, appreciable
quantities of both passed through the filters, with
rejections of u = 0.95 and u = 0.985 respectively.
This suggests that the concentration-polarization
layer formed during serum filtration serves to seal
the membranes, which are otherwise slightly
permeable to these proteins. We have assumed that
basement membranes are impermeable to
molecules as large as IgG in using this molecule as
a ‘leak detector’; but the membranes may be very
slightly permeable to IgG, this permeability being
masked by polarization effects when serum, which
contains larger proteins, is filtered.
It is difficult to assess the relevance of these
studies to glomerular filtration in vivo. The
glomerular filtration barrier is complex and there
has been debate as to whether the basement
membrane is the major, or indeed the only, barrier
to filtration (Farquhar, - 1975). Isolated renal
basement membranes in uitro exclude macromolecules progressively as molecular size increases,
indicating that these membranes can act as
molecular sieves (Robinson & Brown, 1977). However, the presence of additional barriers to filtration
in vivo could have important consequences
(Rodewald & Karnovsky, 1974). Our membrane
filter pads consisted mainly of multiple layers of
isolated tubular rather than glomerular basement
membranes. This experimental model could be
improved by using much thinner membrane pads
comprising glomerular membrane but, for the filter
pads to function as molecular sieves, they need to
be free from leaks. This can only be achieved so
far by constructing multiple layers from fragments
of flat tubular membrane, which seal by overlapping.
The filtration pressure of 150 kPa provided
reasonable flow rates; as the membrane pads were
between 20 and 40 membrane layers thick (Robinson & Brown, 1977), the pressure fall across each
membrane (3-7 kPa) was close to the physiological filtration pressure. At a lower filtration
pressure of 50 kPa, the membranes still showed
polarization effects when serum albumin was
filtered; rejections were slightly lowered, presumably as convective flow through the membranes fell
while diffusive flow was maintained. At very low
filtration pressures the membrane pads were leaky,
presumably as some pressure was required to
maintain an effective seal between the membrane
fragments.
Our very high rejections for serum proteins are
118
T. G. Cotter and G. B. Robinson
similar to those reported for glqmerular filtration in
vivo (Renkin & Gilrpore, 1973), and the hydraulic
permeability of our membrane filter pads (270 k
50 nl rnin-' mm-2 for serum) falls within the
range calculated for the glomerplar filtration rate in
vivo (Renkin & Gilmore, 1973). These similarities
suggest that the model in vitro reflects some of the
characteristics of glomerular filtration in vivo, and
our results show that the phenomenon of concentration-polarization profoundly affects the
behaviour of these basement membrane filters, as it
does for synthetic microporous membranes. We
presume that this general phenomenon also affects
ultrafiltration processes in vivo, and there is
evidence that it may influence glomerular filtration.
Stirring in glomerular capillaries probably results
from the sweeping effects of the erythrocytes as
they flow through the capillary lumen (Deen et al.,
1974). Thus a fall in blood velocity would diminish
stirring so that concentration-polarization effects,
with reduced hydraulic permeability and increased
permeation of macromolecules, should become
apparent. Micropuncture studies show that single
nephron filtration rates fall as the glomerular
plasma flow rate decreases under conditions where
changes in filtration pressures are minimal
(Robertson, Deen, Troy & Brenner, 1972), as is
predicted by polarization theory. Again as predicted, the apparent permeability of the glomerular
barrier to macromolecules is decreased when
plasma flow, and hence stirring efficiency, is
increased (Gassee, 1973; Chang, Ueki, Troy,
Deen, Robertson & Brenner, 1975). In renin- or
angiotensin-induced proteinuria a fall in glomerular
filtration rate and a rise in protein permeability
across the filtration barrier occur (Pessina & Peart,
1972). As angiotensin may increase the resistance
of the efferent arterioles (Pessina, Hulme & Peart,
1972), this would reduce blood velocity in the
glomerular capillaries, so permitting polarization
effects to become pronounced. Micropuncture
studies of the effects of angiotensin suggest that
concentration-polarization could explain the alteration in glomerular function, as glomerular
plasma flow fell, along with a decreased rejection of
dextran and dextran sulphate (Bohrer, Deen,
Robertson & Brenner, 1977; Brenner, Bohrer,
Baylis & Deen, 1977). In these experiments the
predicted fall in glomerular filtration rate was offset
by an increase in filtration pressure.
We do not know whether concentrationpolarization effects contribute to pathological
proteinuria in humans. Studies of nephrotoxic
serum nephritis in rats indicate that proteinuria is
accompanied by a decreased glomerular permeability towards water, as evidenced by reduced
glomerular filtration rates (Von Baeyer, Van Liew,
Klassen & Boylan, 1976), or by increased filtration
pressures with unchanged filtration rates (Bennett,
Glassock, Chang, Deen, Robertson & Brenner,
1976; Chang, Deen, Robertson, Bennett, Glassock
& Brenner, 1976). Thus polarization effects could
be important and we suggest that they should be
taken into consideration in studies of normal and
abnormal glomerular function.
Acknowledgments
T.G.C. is grateful to the Trustees of the Pirie-Reid
Fund, the University of Oxford, for providing a
scholarship for postgraduate study. We gratefully
acknowledge the technical assistance of Ms
Jennifer Byrne.
References
BENNETT,C.M., GLASSOCK,
R.J., CHANG,R.L.S., DEEN,W.M.,
ROBERTSON,
C.R. & BRENNER,B.M. (1976) Studies of
experimental glomerulonephritis in the rat using dextran
sulphate. Journal of Clinical Investigation, 51, 1287-1294.
BLATT, W.F., DRAVID,A., MICHAELS,AS. & NELSEN,L.
(1970) In: Membrane Science and Technology, pp. 47-97.
Ed. Flynn, J. E. Plenum Press, New York.
BRENNER,
B.M., BOHRER,P., BAYLIS,C. & DEEN,W.M. (1977)
Determinants of glomerular permselectivity : Insights derived
from observations in vivo. Kidney International, 12, 229237.
BOHRER,M.P., DEEN,W.M., ROBERTSON,
C.R. & BRENNER,
B.M. (1977) Mechanism of angiotensin 11-induced proteinuria
in the rat. American Journal of Physiology, 2, ~ 1 3 - ~1. 2
CHANG,R.L.S., DEEN, W.M., ROBERTSON,
C.R., BENNETT,
B.M. (1976) Studies of
C.M., GLASSOCK,
R.J. & BRENNER,
experimental glomerulonephritis in the rat using neutral
dextran. Journal of Clinical Investigation, 51,1271-1286.
CHANG,R.L.S., ROBERTSON,
C.R., DEEN,W.M. & BRENNER,
B.M. (1975) Permselectivity of the glomerular capillary wall
to macromolecules: I. Theoretical considerations. Biophysical Journal, 15,86 1-886.
CHANG, R.L.S., UEKI, I.F., TROY, J.L., DEEN, W.M.,
ROBERTSON,
C.R. & BRENNER,
B.M. (1975) Permselectivity
of the glomerular capillary wall to macromolecules: 11.
Experimental observations in the rat. Biophysical Journal,
159 887-906.
DEEN,W.M., ROBERTSON,
C.R. & BRENNER,
B.M. (1974) Concentration-polarisation in an ultrafiltering capillary. Biophysical Journal, 14,4 12-43 1.
DEJMEK, P. (1975) Separation processes by membranes, ion
exchange and freeze concentration in food industry.
International Symposium, Paris, A2, 1.26.
DUBOIS,R., DECOODT,P., GASSEE,J.P., VERNIORY,
A. &
LAMBERT,J.P. (1975) Determination of glomerular intracapillary and transcapillary pressure gradients from sieving
data: I. A mathematical model. P'ugers Archiv, 356, 299316.
FARQUHAR,M.G. (1975) The primary filtration barrier, basement membrane or epithelial slits? Kidney International, 8,
197-2 I 1.
GASSEE,J.P. (1973) Effect of acetylcholine on glomerular
sieving of macromolecules. P'ugers Archiv, 342,239-254.
Filtration through basement membranes
GOLDSMITH,
R.L. (197 1) Macromolecular ultrafiltration with
microporous membranes. Industrial and Engineering
Chemistry (Fundamental), 10, 113-120.
LIGLER,F.S. & ROBINSON,G.B. (1977) A new method for the
isolation of renal basement membrane. Biochirnica et
Biophysica Acta, 468,327-340.
LOWRY,O.H., ROSEBROUGH,
N.J., FARR,A.L. & RANDALL,
R.J. (1951) Protein measurement with the Folin phenol
reagent. Journal ofBiologica1 Chemistry, 193,265-275.
PESSINA,A.C., HULME, B. & PEART, W.S. (1972) Renin
induced proteinuria and the effects of adrenalectomy: 11.
Morphology in relation to function. Proceedings of The Royal
Society, London, series B, 180,61-71.
A.C. & PEART,W.S.(1972) Renin induced proteinuria
PESSINA,
and the effects of adrenalectomy. Proceedings of The Royal
Society, London, series B, 180,43-60.
RENKIN,E.M. & GILMORE,J.P. (1973) In: Handbook of
Physiology, vol. 8, pp. 185-248. American Physiological
Society, Washington.
119
ROBERTSON,
C.R., DEEN,W.M., TROY,J.L. & BRENNER,
B.M.
(1972) Dynamics of glomerular ultrafiltration in the rat: 115.
Hemodynamics and autoregulation. American Journal of
Physiology, 223,1191-1200.
ROBINSON,
G.B. & BROWN,R.J. (1977) A method for assessing
the molecular sieving properties of renal basement membranes in vitro. FEBS Letters, 18, 189-193.
RODEWALD,R. &. KARNOVSKY,
M.J. (1974) Porous substructure of the glomerular slit diaphragm in the rat and
mouse. Journal of Cell Biology, 60,423-433.
RYAN, G.B. & KARNOVSKY,
M.J. (1976) Distribution of
endogenous albumin in the rat glomerulus; role of hemodynamic factors in glomerular barrier function. Kidney
International, 9,36-45.
VON BAEYER,H., VAN LIEW, J.B., KLASSEN,J. & BOYLAN,
J.W. (1976) Filtration of protein in the anti-glomerular basement membrane nephretic rat: a micropuncture Study.
Kidney Internat ional, 10,425-43 7.