Clinical Science (1970) 39, 705-724.
EXTRAVASCULAR ALBUMIN MASS AND EXCHANGE
IN RAT TISSUES
J. KATZ, G. BONORRIS, SYBIL GOLDEN
AND
A. L. SELLERS
Medical Research Institute, Cedars-Sinai Medical Center, Los Angeles, California
(Received 23 March 1970)
SUMMARY
1. Extravascular albumin in carcass, skin and gut of rats was extracted and the
albumin content estimated by several methods. Assay by electrophoresis on acrylamide gel, by immunodiffusion and by radioimmunoassay were in essential agreement. The method used previously, precipitation with antibody followed by alcoholTeA extraction, underestimates the amount of albumin in tissue extracts, because
extraction from the antibody precipitate is not complete. This method is valid,
however, for specific activity determination.
2. Normal rats contain from 500 to 650 mg of albumin per 100 g body weight.
Of this, 20-25% is in the circulation, 3~0% in the carcass (mainly but not exclusively muscle), 20-25% in skin and 10% in gut.
3. The extracellular water of muscle, carcass, skin and gut was estimated from the
distribution of mannitol and sulphate. With the exception of gut, both methods
agreed closely. Extracellular, extravascular water constitutes about 23% of the body
weight of 150-200 g rats. The extracellular water in muscle is about 20% and in skin,
40%. In gut the extracellular water cannot be estimated reliably by these compounds.
4. Muscle contains about 3·5 mg/g of extravascular albumin; skin and gut, 7-8
mg/g. The concentration ofextravascular albumin in extracellular water of muscle and
skin is 16-20 mg/ml, or 50-60% of the concentration in plasma. In the small intestine
the concentration of albumin is higher, possibly similar to that in plasma.
5. In rats with severe aminonucleoside nephrosis, body albumin was depleted to
100-200 mg/IOOg. Of this, 15-25% was in plasma, 50% in carcass, and about 15% in
skin. Ascitic fluid contained only a few mg of albumin.
6. The specific activity of extravascular albumin of tissues was followed after
intravascular injection of 125 1_ or 1311-labelled albumin. The specific activity of
carcass albumin increases rapidly, becoming equal to that in plasma after less than
Correspondence: Dr Joseph Katz, Cedars-Sinai Medical Research Institute, 4751 Fountain Avenue, Los
Angeles, California 90029, U.S.A.
705
706
J. Katz et al.
2 days. The specific activity of albumin in skin increases much more slowly and
becomes equal to that of plasma after 4 days. Labelling of albumin of gut is even
slower. The specific activities in tissue never exceed that in plasma.
7. In severely nephrotic rats, specific activities in carcass and skin become equal
to that in plasma within 2-3 days and remain equal thereafter. Specific activity of
albumin in ascitic fluid increases to reach values as much as sixteen-fold those in
plasma.
8. The extravascular pool, as calculated by multicompartmental analysis from the
slopes and intercepts of the plasma curve, is about equal to that in plasma and in
severely nephrotic rats is less than that in plasma. Discrepancies between calculated
and observed extravascular albumin masses is by a factor of 3 in normal rats and 10 or
more in severely nephrotic rats.
9. Specific activity of extravascular albumin as calculated from multicompartmental analysis is 1·5 times that in plasma in normal rats and at least six times that in
plasma in severely nephrotic rats. Actually, the specific activities in extravascular and
vascular albumin ultimately become and remain equal.
10. It is concluded that the multi compartmental model of vascular pool exchange
with one or two extravascular pools is not valid for rats and probably not for other
animal species.
The distribution of plasma albumin in blood and the extravascular space, and its exchange
between these compartments, is of considerable physiological interest. Albumin distribution
has been calculated by several isotopic methods, and it has been estimated that in most species
from 50 to 60% of body albumin is extravascular (Schultze & Heremans, 1966; Wetterfors,
1965b). We have observed the remarkable stability of circulating albumin in rats despite
massive rapid losses of albumin from renin-induced proteinuria (Katz, Sellers & Bonorris,
1964). Such stability was compatible with a much larger albumin mass than had been indicated
from isotopic studies. This prompted us to measure extravascular albumin directly by extraction from tissues. We have found (Sellers, Katz, Bonorris & Okuyama, 1966; Katz, Sellers &
Bonorris, 1963; Sellers, Katz & Bonorris, 1968; Katz, Sellers, Bonorris & Golden, 1970) that
in normal and aminonucleoside-nephrotic rats, the mass of extravascular albumin is three
times that in blood. We have also found that after equilibration following injection of radioactive iodine-labelled albumin, the specific activity of extravascular albumin became equal to
that in plasma, whereas according to conventional models of plasma protein metabolism the
specific activity of extravascular albumin in the rat should be considerably higher than that in
blood. Our findings, if substantiated, would be inconsistent with currently held concepts of
plasma protein metabolism based on tracer methods, and therefore a careful examination of
our methods was called for.
We had previously determined extravascular albumin in tissue extracts by a procedure
combining precipitation with specific antibody and extraction with alcoholic trichloroacetic
acid (antibody-alcohol-TCA method). This procedure has rarely been used, and comparison
with other methods was in order. We find that the antibody-alcohol-TCA method actually
underestimates extravascular albumin. We also report kinetic studies on the exchange of
circulating albumin with that in the extracellular fluid of muscle, skin and gut, and show that
the exchange kinetics between circulating albumin and that in these three tissues differ greatly.
Extravascular albumin in the rat
707
METHODS
Materials
The preparation of rat albumin and rabbit anti-rat albumin antibody has been described
(Sellers et al., 1966, 1968). Albumin was iodinated with 1311 or 1251 as previously described
with ICI and used after screening. In later work it was iodinated electrolytically (Katz &
Bonorris, 1968) and used without screening.
Extraction of extravascular protein of tissues
The procedure was essentially as described. To correct for residual blood in tissues,
rats were injected i.v. with 0'1-0·2 ,uCi 125 1_ or 1311-labelled albumin. Rats that were
previously injected with rat [1251]albumin were injected with human commercial p311]albumin.
Three to 6 min later the rats were anaesthetized with ether, shaved, the abdomen was opened,
and as much blood as possible aspirated from the aorta. In later experiments, to deplete residual
blood still further, 20-30 m1 of saline was injected into the aorta and the fluid re-aspirated.
The body was either used whole or divided into carcass (skinned, eviscerated body), skin,
gastrointestinal tract (excluding colon) and other viscera (heart, lungs, testes or ovaries, spleen,
liver, kidneys). The whole body or carcass or skin were ground in a meat grinder and 25 g
extracted in a high-speed blender with 250 m1 of ice-cold saline containing 0·05% sodium
, desoxycholate at pH 8·0. Gut and viscera were extracted directly in the blender without prior
grinding. The extracts were immediately centrifuged in the cold and the turbid supernatants
frozen. For analysis the extracts were thawed, cleared immediately by centrifugation and
promptly analysed. This is essential with gut extracts to avoid breakdown ofprotein by digestive
enzymes.
Determination of albumin in tissue extracts
The albumin content of the extract was determined by disc electrophoresis on acrylamide gels
and by three different immunochemical methods: (a) the antibody-alcohol-TCA method used
previously, (b) radioimmunoassay and (c) immunodiffusion.
Corrections were made for albumin contained in residual blood. The occluded blood albumin
was calculated from the activity in tissue extracts and specific activity of blood albumin. When
animals were previously injected with radioactive iodine-labelled albumin, a double isotope
assay for 1251 and 131 1 was used.
Disc electrophoresis. Tissue extracts were electrophoresed on acrylamide gel, stained with
Aniline Black, and the gels scanned using the Canalco (Bethesda, Maryland, U.S;A.) materials
and apparatus. Albumin content of the extract was calculated by two procedures: (a) the
fraction ofthe total area in the albumin peak was obtained by scanning, and this was multiplied
by total protein content of the extract determined by the biuret reagents according to Lowry,
Rosebrough, Farr & Randall (1951) and Kingsley (1942); (b) albumin content was calculated
from the peak area by use of a reference curve, obtained by plotting peak areas of a series of rat
albumin standards. The standard curve when extrapolated passed through the origin and was
linear for albumin concentrations up to 60 ,ug per disc.
Antibody-alcohol-T'CA method. Samples of the extract containing from 150 to 300 ,ug of
albumin were incubated overnight with excess antibody at 37°. The precipitate was washed and
extracted with 2-3 ml of 5% trichloroacetic acid in 90% ethanol (very little albumin or radio
B
708
J. Katz et al.
activity was removed by a second extraction). Samples of the TCA extract were analysed for
protein according to Lowry et al. (1951) and, if specific activity was measured, assayed for 1311
or 1251 in a gamma scintillation counter.
Radioimmunoassay. Rat [1251]albumin of high specific activity (10-50 x 106 cpm/mg)
was prepared by small-scale electrolytic iodination (Katz & Bonorris, 1968). For dilution, a
0·05 Mborate buffer, pH 8·5, 0·1 Min NaCI, and containing 10% of dog or human serum, was
used. The albumin was diluted so that 0·05 ml contained about 10000-15000 cpm, Antibody
to rat albumin was diluted so that 0·05 ml precipitated 3 /lg of albumin.
Samples containing 5-15 /lg of albumin were added to small test tubes and made up to 0·20
ml with borate buffer. After addition of 0·05 ml each of the [1251]albumin and antibody solution, the tubes were incubated overnight at 4°. One-half ml of 64% saturated ammonium sulphate was added and after 1 h the tubes were centrifuged. One-halfml of the supernatant was
assayed for 1251. Standards containing from 4 to 16 /lg of albumin were processed simultaneously. The curve is linear in this range but departs markedly from linearity at lower and higher
concentrations.
Immunodiffusion. Immunodiffusion was performed according to Mancini, Carbonara &
Heremans (1965) in agarose gels. Diffusion was for 24 h. The precipitate areas were projected
with a photographic enlarger and their diameters measured. Plots of albumin concentration
versus the square of the diameter were linear.
Extracellular water
Rats were anaesthetized with ether, bilaterally nephrectomized, and injected i.v. with 0·5 ml
of 0,9% NaCl containing about 10 /lCi Na235S04 and 50 /lei mannitol-I-T (New England
Nuclear, Boston, Mass., U.S.A.). 30 to 45 min later the rats were injected i.v, with 0,2 /lCi
of human [1311]albumin. Three or 4 min later the rats were bled, the tissues separated and
ground and extracted as described above. Samples of the extracts were treated with 1/4
volume of 10% AgN0 3 in 50% trichloroacetic acid and the supernatants clarified by high-speed
centrifugation. This treatment removed all the 1311. Plasma and blood were similarly treated.
The volume of residual blood in the tissues was determined from the 1311 count in extracts and
in blood. The 35S and tritium content of the AgN0 3-treated extracts were assayed in a twochannel liquid scintillation counter. The sulphate space was calculated as follows:
_ 35S cpmjg tissue-ml residual blood x 35S cpmjml blood
Su1ph ate space 35S cpm jml p 1asma
Plasma was taken as 100% water, neglecting the partial volume of the dissolved proteins.
Mannitol space was calculated in exactly the same manner from the tritium counts. The sulphate
or mannitol spaces were expressed as mlj100 g of blood-free tissue. Preliminary experiments
indicated that the distribution space of these two compounds approached a plateau and
remained unchanged from t to 1 h after injection.
Other methods
Albumin in plasma was also determined by the alcohol-TCA method of Debro, Tanner &
Korner (1957) and total protein according to Lowry et al. (1951). The induction of aminonucleoside nephrosis and other procedures were as previously described (Sellers et al., 1966,
1968).
Extravascular albumin in the rat
709
RESULTS
In Fig. 1 scans of acrylamide discs prepared from rat plasma and desoxycholate extracts of
carcass, skin and gut are reproduced. Total extractable protein as determined by Lowry's
method was 35-50 mg/g in carcass, 25-40 mg/g in skin, and 50-70 mg/g in gut. The apparent
protein content of tissue was dependent on the method of analysis. Thus in one gut extract
J
NORMAL SERUM
\:60
43%
CARCASS
1:11
52%
SKIN
1:11
54%
46%
FIG. 1. Acrylamide disc electrophoresis of serum and tissue extracts of normal (left) and aminonucleosidenephrotic (right) rats. The volume of diluted serum or extracts applied to discs was
0·075 mI. Dilution of tissue expressed as mI/g. The time of electrophoresis differed. Scanning was
done at the same sensitivity for all discs. Percentage of total area under the curve in the albumin
peak is indicated.
apparent extractable protein was 65 mg/g as determined by the Lowry method, but 45 mg/g
with the Kingsley biuret reagent using rat albumin as a standard.
The electrophoretic patterns revealed the presence of plasma proteins as well as protein
derived from tissue. Estimation of the albumin content of tissue extracts by multiplying total
apparent protein content by the fraction in the albumin peak led to unreasonably high values.
When albumin was estimated from the area under the albumin peak, using as reference a
standard curve obtained from electrophoresis of rat albumin, lower values were obtained
which were in good agreement with those measured by a number of other methods (see below).
In plasma both methods of estimation gave identical results. In gut extracts, on the other hand,
the apparent albumin content was 20-30 mg/g by the first and 7-10 mg/g by the second procedure. It appears that tissue extracts contain peptidic material which yields colour with the
Lowry and biuret reagents but which either does not stain or is washed out from the acrylamide discs. This fraction is present in all tissue extracts and is very prominent in that from gut.
J. Katz et al.
710
The albumin content of plasma and tissue extracts as determined by three methods-(l)
from the albumin peak on acrylamide gel discs, (2) by radial immunodiffusion, and (3) by radioimmunoassay-are compared in Table 1. Values from each of these methods are in essential
agreement. The purpose of this comparison was not for a statistical evaluation of the precision
TABLE 1. Methods for determination of albumin in plasma and tissue extracts
Condition
Tissue
Acrylamide
gel
Immunodiffusion
TCARadioimmunoassay alcohol
(mg/ml)
Normal
Mild nephrotic
Severe nephrotic
Serum
Serum
Serum
34'0
17'2
31·5
16'7
5'0
32·0
5-8
30·0
17·9
10·6
(pgJml of extract)
Normal
Normal
Normal
Normal
Normal
Normal
Normal
Nephrotic
Nephrotic
Nephrotic
Nephrotic
Nephrotic
Relative values
Total body
Total body
Carcass*
Skin*
Gut*
Carcass
Skin
Carcass]
Skint
Total body
Carcass
Skin
415
450
365
600
140
150
210
97
463
483
427±23
750±34
830±29
438
696
178± 10
304±29
107
141
240
100
714
810
455± 12
750±18
825±30
428
807
165± 8
300±17
92
126
238
98
612
873
432
258
348
* Mean and SEM of six rats.
t Mean and SEM of fifteen rats.
Each value for tissues by radioimmunoassay or acrylamide gel expressed as percentage
of that by immunodiffusion, which was set as 100%. and values averaged.
or accuracy of these procedures; it was rather to show that the methods based on diverse
principles are in reasonable agreement, usually within 10-15% of each other for individual
samples, The agreement holds for all tissues and under widely divergent conditions. Agreement
between the electrophoretic method and those based on binding to antibody strongly supports
the conclusion that these three methods represent a true measure of plasma albumin in tissues.
Of the three methods, the radial immunodiffusion assay is by far the easiest, most economical
and most reproducible, and is the method of choice.
In Table 1 results obtained by the alcohol-T'CA method of Debro et al. (1955) are given.
This method has commonly been used for albumin assay. For a large series of normal plasma
samples, we found that results agree very closely with other procedures, but the values
for tissue extracts are twice those obtained by the three other methods shown. It is well
known (Munro & Downie, 1964) that there are cellular proteins soluble both in water and in
Extravascular albumin in the rat
711
alcoholic TCA, and therefore the procedure is not suited for tissue extracts. Excessivealbumin
values were also obtained for sera of severely nephrotic rats (Table 1) and for sera from haemolysed blood. Thus this method is not suitable for abnormal sera and it must be used with
caution.
In the immunochemical assays the specificity of the antibody is critical. When tested by
immunoelectrophoresis on agarose gels against rat plasma, there was, in addition to the
pronounced albumin arc, a faint second arc between the C( and f3 globulin region. With tissue
extracts the second arc could only be detected if the strips were overloaded with protein.
We compared our antibody to two other rat albumin antibody preparations, a commercial
preparation (Cappel Laboratories, Inc., Downington, Pa., U.S.A.) and one obtained through
the courtesy of Dr D. A. Darcy (Chester Beatty Research Institute, Institute of Cancer Research,
Royal Cancer Hospital, Belmont, Sutton, Surrey). The Darcy antibody showed only a single
albumin arc on immunoelectrophoresis, but the commercial preparation (Cappel) had two
additional minor arcs. In Table 2 results of albumin assay of tissue extracts by immunodiffusion
TABLE
2. Comparison of three antibody preparations in the analysis of albumin
by immunodiffusion in plasma and carcass extracts
Antibody source
Fraction
(No. of rats)
Normal plasma (4) (mg{mI)
Extract, normal rat (4) (pg{ml)
Extract, nephrotic rat (2) (pg{ml)
Present work
Darcy
Cappel
2%±O'8
498±17
157
29'9±1'8
500±18
145
28'7±1'4
490±23
173
Mean and SEM.
and radioactive immunoassay are compared for the three antibody preparations. Results
obtained by all three are virtually identical.
The isolation of tissue albumin employed in previous studies was based on two properties: .
(a) precipitation with specific antibody, and (b) solubility of precipitated albumin in alcoholTCA. It was felt at that time that a rigorous method, employing the double criteria, was to be
preferred even if it led to underestimation of tissue albumin. In Table 3 the albumin content
of plasma and tissue extracts, as assayed by the antibody-alcohol-TCA and the immunodiffusion methods, are compared. In plasma the antibody-alcohol-TCA method gives values
5-10% less than the immunodiffusion procedure. In carcass extracts, however, this method
underestimates albumin content by as much as 3~O%. The main reason for this underestimation is shown to be incomplete extraction of albumin from the antibody precipitate (Table 4).
Most of the albumin in plasma (78%) was extracted from the precipitate, but with carcass
extracts a substantial fraction could not be solubilized even by exhaustive extraction with
alcohol-TCA. Peters (1958) also reported incomplete extraction of albumin from antibody
precipitates.
While the recovery of tissue albumin by antibody precipitation and alcohol-TCA extraction
is incomplete, the method provides a valid measure of the specificactivity of albumin in tissues.
When specific activities of tissue albumin were determined by this method and also calculated
J. Katz et al.
712
TABLE 3. Comparison of the antibody-alcohol-TCA and immunodiffusion methods for estimation of tissue albumin
Tissue
(No. of rats)
Antibodyalcohol-TCA
Immunodiffusion
Ratio
(mg/ml)
Plasma (10)
29·3
(27'3-32'1)
32·8
(31'5-34'0)
0·89
(mg/g)
Carcass (8)
Muscle (4)
Skin (8)
Stomach and small intestine (2)
Small intestine (6)
2·2
(2,0-2'9)
2-4
(2-2-2-6)
5·1
(4'3-5'7)
6·9
7·0
(6'2-8'1)
3-8
(3,2-4'8)
0'58
4-4
0'55
(3·8-5·0)
8'3
(7-8-9'3)
8·7
0'62
7-9
0·89
(7'0-9,2)
Mean and ranges in parentheses. Not corrected for residual
vascular albumin, which is 0·2--0·6 mg/g.
TABLE 4. Extraction of radioactive iodine-labelled albumin with alcohol-TCA
Distribution of activity (%)
Tissue
Plasma
Carcass
Skin
Small intestine
Total 131I
activity
TCA precipitate
(cpm/g)
TCA TCA-alcohol
soluble*
soluble
272000
36000
65000
61000
5·5
16·0
7·6
11·0
94
86
88
91
Antibody precipitate
Antibody
soluble
TCA-alcohol
Soluble
Insoluble
78
47
68
77
11
36
12
10
11
25
19
14
Mean of two rats,S days after injection of 131I_albumin.
* There was negligible activity in the residue.
from the amount of total protein activity and albumin content determined by immunodiffusion,
similar values were obtained (Table 5). This table also shows that specific activities of carcass
and skin albumin are equal to those of plasma 4-5 days after the injection of radioactive iodinelabelled albumin.
Blood albumin in tissues
Some of the albumin in the tissue extracts is due to residual blood. About half of the circulating blood was removed by aspiration from the aorta. The carcass or muscle was found to
713
Extravascular albumin in the rat
contain about 0·02 ml/g of residual plasma, and skin and gut, 0·01-0·015 ml/g. The blood
content of the pooled viscera (excluding gut) was variable but quite high, and blood albumin
constituted up to half of the total albumin in these extracts. Injection of the rats with saline
(see Methods), followed by aspiration, considerably reduced residual blood, bringing corrections for muscle to less than 10% and making them insignificant for skin and gut.
TABLE 5. Determination of specific activity of tissue albumin
Days after
injection of
radioactive
iodine-labelled
albumin
Antibodyalcohol-TeA
Immunodiffusion
(cpm/mg)
co
(cpmjmg)
(%)
4*
Plasma
Carcass
Skin
Small intestine
7600
100
101
100
90
8070
100
104
102
93
5*
Plasma
Carcass
Skin
Small intestine
10600
100
107
98
99
11400
100
106
90
87
* Mean of two rats.
Extravascular albumin content of tissues
Table 6 summarizes our findings on the distribution of extravascular albumin. Results are
presented as mg/g of blood-free tissue and percent of total body albumin. Earlier results were
obtained by the antlbody-alcohol-TCe, method which, as shown above, underestimates
extravascular albumin. The previously published values and those of experiments 1-4 were
revised, using correction factors from the last column of Table 3. In later work (experiments 5
and 6), the immunodiffusion method was used.
Extravascular albumin ranges from 400 to 500 mg/l 00 g and constitutes 75-80% of total body
albumin. The albumin content ofcarcass is one and a halftimes that in blood, and this is mainly
but not exclusively in muscle. Substantial amounts of albumin are likely to be present in the
extracellular fluid of connective tissue, tendons, lymph vessels, bone marrow, etc. The albumin
content of skin equals that in plasma and gut contains approximately half that in plasma. The
extravascular albumin of the pooled viscera was determined only in a few experiments and it
constitutes less than 10% of total body albumin. The albumin content of the various visceral
organs may differ considerably.
In experiment 2 of Table 6 the plasma decay of the albumin specific activity was measured
(curve of Fig. 2 below) and extravascular albumin was calculated from the isotopic data.
The ratio of plasma albumin to total body albumin is expressed according to Nosslin (1964) as
Plasma pool
Body pool
~[cdbd2
~cj/bj2
419
(75%)
139
(25%)
558
364
(73%)
133
(27%)
497
61
127
143
12
141
(110-172)
Exp. 3
417
(74%)
147
(26%)
564
66
131
182
8
194
(172-210)
Exp.4
136
(26%)
170
(31%)
130
(24%)
67
(13%)
425
Mean
expts 1-4
496
(80%)
124
(20%)
620
-
-
-
8
157
(120-190)
Exp. 5
220
(37%)
144
(24%)
58
(10%)
464
(78%)
138
(22%)
602
7
146
(105-160)
Exp. 6
Experiments 1-4 performed over a 2-year period with the antibody-alcohol-T'Ca method and results corrected by using
factors from column 3 of Table 3. Experiments 5 and 6 performed by immunodiffusion. In experiments 1-4 and experiment 6,
total extravascular albumin was estimated assuming that the carcass, skin and gut contain 90% of total extravascular albumin.
Total
Blood
510
(80%)
130
(20%)
640
412
(76%)
133
(24%)
545
Total EV
73
76
-
Gut
138
125
-
Skin
170
18
144
(1l0-160)
14
141
(124-152)
173
40
122
(100-145)
Exp.2
Exp. 1
-
Albumin (mg/l00 g) in:
Carcass
No. of rats
Animal weight (g)
Sellers
et al, (1966)
TABLE 6. Distribution of albumin in the rat body
:-
~
.....
~
~
.....
N
~
oj::>.
......:J
.......
Extravascular albumin in the rat
715
where the c's and b's are the intercepts and slopes of the exponential plasma curve. The intercepts of the semilogarithmic plasma curve in this experiment were 0·28 and 0,72, and the slopes
0·013 and 0·10 (fraction of dose per h). Thus the ratio is
0,72 0·28 J2
Plasma pool
Total pool
= [ o:tO +Q.0i3
= 0.48
0·72 0·28
O' I 02 0'0132
The curve of Fig. 3 can actually be resolved into three exponential components, indicating a
large and a small extravascular pool, which turns over very rapidly. The plasma albumin curve
obtained from individual rats is usually resolved into two exponential components and only
occasionally is a third component noted (Katz et al., 1963). The evaluation of the third component is subject to a large experimental error and requires careful sampling during the first
hours after injection. Consideration of a third component has an insignificant effect on the mass
of total extravascular pool and the component is neglected. Calculation thus gives a value of
48% of total body albumin in circulation and 52% in extravascular space. In previous experiments (Katz et al., 1963)the calculated values for plasma albumin by this method ranged from
40 to 55% in circulation. Thus by the isotopic method extravascular albumin is calculated to
be some 140 mg/IOO g as compared to 420 mg by direct extraction.
Concentration of albumin in tissue fluid
The extravascular albumin content of skin and gut was twice that of muscle. It seemed likely
that this would be related to the extracellular water content of the tissues.
The mannitol and sulphate spaces of total body, carcass, muscle and skin agreed closely
(Table 7). The extracellular water in tissues represented 25% of the blood-free body weight.
Including blood, extracellular fluid constitutes 28-29% of (total) body weight in lean rats
of l5D-200 g. Swan, Modisso & Pitts (1954)found in dogs that the sulphate and mannitol spaces
were identical and constituted 23% of body weight. The extracellular water content of rat
muscle was 20%, somewhat less than that in the total body but the extracellular water of skin
was much higher, 40% of the blood-free organ. Total water content of blood-free body and
of skin, as determined by desiccation, were similar and equalled 73-78%; thus extracellular
water constitutes over half of total skin water.
The apparent sulphate space of gut was rather variable but greater than the mannitol space
and frequently exceeded the total volume of that tissue. Rinsing out of the intestines removed
most of the sulphate and one-fourth to one-third of the mannitol (Table 7). It appears that
sulphate is excreted into the lumen, but this does not altogether account for the large apparent
sulphate space. It may be that in the gut sulphate penetrates the cells or exchanges with sulphate of mucopolysaccharides. The apparent mannitol space of rinsed gut was also quite
variable (16-27%). The determination of the extracellular water of stomach and intestine is thus
not valid with sulphate and not certain with mannitol.
In Table 8 the distribution of albumin in tissues and the concentration of albumin in the
extracellular water of tissues are presented. The concentration of albumin in extravascular,
extracellular water of muscle and skin is similar, but since the extracellular water content of
skin is twice that in muscle, the total albumin content of skin is high. With the exception of gut,
the .concentration of albumin in extracellular water is quite uniform, equalling 50-55X that
J. Katz et at.
716
TABLE 7. Sulphate and mannitol space of tissue
Sulphate
Mannitol
(% of blood-free mass)
Total body
Carcass
Muscle
Skin
Stomach and small intestine:
Unrinsed
Rinsed
27
(26-29)
24
(22-26)
20
(17-22)
39
(37-42)
26
(23-28)
23
(19-26)
20
(17-22)
40
(34-42)
125
(69-160)
44
(40-48)
30
(25-41)
22
(16-27)
Means and ranges (in parentheses) from eight rats
except for intestine, where there are four animals in
each group.
TABLE 8. Distribution of extravascular albumin and concentration in extracellular
water
Tissue
Albumin
Fraction of
body weight* Content Distribution
(mg/g)
co
co
Blood
Plasma
Carcass
Extracellular water
Amountj Albumin cone.
(ml/g)
(mg/ml)
8
(7-8)
33
58
(55-62)
3·5
24
35
16
0·20
19
18
(15-21)
7-9
26
0·40
20
Gut
6
(5-7)
8·2
9
0·2-0·3
25-35
Other viscera
10
(9-12)
0·26
18
Muscle
Skin
3·9
0·23
Blood-free body
Total
6
4·6
100
100
* Rats weighing from 140 to 210 g.
t Per g blood-free tissue.
Extravascular albumin in the rat
717
in plasma. The concentration of albumin in rinsed gut cannot be estimated with precision,
since the extracellular water content is uncertain and since as much as 20% ofthe albumin may
be in the lumen. However, a conservative estimate is 25 mg/ml, and it may even be equal to the
concentration in plasma (30-35 mg/ml). This high albumin content is of interest and warrants
further studies.
The concentration of albumin in lymph is likely to represent the average concentration of
albumin in the extravascular fluid it drains. Numerous determinations for albumin content of
the lymph for many animal species are available (Altman & Dittmer, 1961) and the values are
quite variable, ranging from 40 to 70% that in plasma for thoracic and cervical lymph and 80100% that in plasma for hepatic and intestinal lymph. Our results (Table 8) are consistent with
50-60% for thoracic lymph and at least 80% for intestinal lymph.
Extravascular albumin in nephrotic rats
The severity of symptoms varied considerably in each group, and this explains the variability
of the data. Rats with severe aminonucleoside nephrosis excrete 10-15 mg albumin h -1 100 g
body weight"? in the urine, and their plasma albumin may decrease from a normal of 3035 mg/ml to as low as 6 mg/ml, They are oedematous and frequently accumulate as much as
10-15 ml of ascitic fluid/IOO g body weight. The albumin concentration of ascitic fluid is low
(0'1-0·2 mg/ml) and thus it constitutes only a small fraction of total body albumin. Total body
albumin is reduced from 600 mg/IOO g to values of 150-200 mg/100 g and occasionally as low
as 100 mg/IOO g (Table 9). These rats are markedly oedematous, and thus albumin content
expressed on the basis of body weight overestimates the extent of albumin loss.
In spite of the large albumin loss, the fraction in circulation did not differ greatly from normal.
There was some indication that the vascular albumin fraction in severe nephrosis is less than
in normal rats, but the data are variable and the conclusion is not statistically warranted.
However, we have observed nephrotic rats with 15% or less of body albumin in the blood,
values not seen in controls. The distribution of albumin between carcass and skin is quite
different from normals, with a greater fraction in carcass and a considerably depressed fraction
in skin.
The kinetics of labelling of extravascular albumin
We have previously shown (Sellers et 01., 1966) that the specific activity of extravascular
albumin became equal to that in plasma 4-6 days after injection of radioactive iodine-labelled
albumin. It appeared of interest to study the course of labelling of individual tissues, and the
results are shown in Fig. 2. The curves are strikingly different. The specific activity of carcass
(mainly muscle) albumin increased very rapidly, equalling that of circulating albumin after
about 12 h. For the next day activity seemed to exceed that of plasma, but thereafter specific
activities of albumin in muscle and in circulating blood became equal and remained equal.
The specificactivity of skin albumin increased for about 24 h, attaining a value approximately
one-third that in plasma, and remained fairly constant for the next 2 days. Due to the rapid
fall in specific activity of plasma albumin, the specific activities became equal on the fifth day
and remained equal thereafter. Labelling of gut albumin was the slowest, and the specific
activity curve indicated complex kinetics.
The plot of the average specific activity of the total extravascular albumin gives the appearance of simple kinetics and does not indicate the existence of its individual constituent pools.
13
Plasma albumin (mg/ml)
166
(124-202)
30
(27-64)
206
(155-257)
225
60
285
Total EV
Blood
Total
100
2
4
9,15
6'7
(%-7'2)
3
188
(182-195)
Severe
Exp. 3
39,65
233,263
88
(72-106)
383
(323-425)
168,224
109
(97-125)
25
(24-27)
84
(73-98)
-
-
-
295
(215-331)
-
-
-
(mg/l00 g) (mg/IOO g) (mg/l00 g)
21
(17-26)
165, 172
Moderate
Mild
146
(II 5-190)
Exp. 3
Exp. 3
100
23
(22-25)
(%)
In experiments 1 and 2 and of Sellers et al. (1968), determinations made with antlbcdy-alcohol-TCa, and results corrected as in
Table 6. Experiment 3 done by immunodiffusion.
149
(119-176)
31
(23-40)
20
(15-27)
11
-
23
(9-40)
-
Gut
120
(90-151)
16
26
(18-33)
31
(22-41)
Skin
-
-
Carcass
47
co
Average
1 and 2
70
(52-88)
(mg/l00 g) (mg/l00 g)
7·9
(6'3-9,2)
180
(160-212)
16
Severe
Exp. 2
98
(55-137)
(mg/g)
168
(132-225)
194
Tissue
18
34
6·8
(5'2-8'8)
Severe
Moderate
No. of rats
Body weight (g)
Exp. 1
State of nephrosis
et al. (1968)
Sellers
TABLE 9. Albumin distribution in tissues of nephrotic rats
......:J
:-
~
(1:>
.....
~
~
~
00
>-'
Extravascular albumin in the rat
719
The mean specific activity of total extravascular albumin attained maximum at about 30 h
after injection and became equal to that in plasma after 5-6 days. The plasma curve in rats can
usually be resolved into two exponential components, consistent with the presence of a single
extravascular pool. Our results show the assumption of such a homogeneous extravascular pool
--Blood
--0-- Carcass
-Ll--Skin
._ .•. _. Gur
2
o
24
48
96
72
120
144
168
Time (h)
FIG. 2. Specific activity of blood and extravascular albumin after the injection of [l 2 s Ilalbumin.
Each point obtained from a single rat.
in the rat to be untenable, and the calculated exchange rate between vascular and extravascular
pools has little meaning.
According to the conventional multicompartmental model of plasma protein metabolism,
the specific activity of extravascular albumin should exceed that in plasma (Schultze & Heremans, 1966; Nosslin, 1964). The ratio of specific activities of extravascular and vascular
albumin is given by the equation
K
S.A. of EV albumin
S.A. of V albumin = K - b i
where b i is the terminal slope of the plasma curve and K is the rate of exchange between extravascular and vascular pools, given by CI b 2 + C2 b.. The equation was first derived by Berson &
Yalow (1957). From the intercepts and slopes of the resolved plasma curve in Fig. 2, the calculated extravascular specific activity should attain a value I-53 times that in circulation. In
a previous experiment (Katz et al., 1963) using individual curves obtained from fifteen normal
rats, an average specific activity ratio of 2-1 (range 1-S-3'0) was obtained. This difference
between theoretical and observed specific activities is highly significant.
J. Katz et al.
720
In nephrotic rats the turnover rate of albumin is greatly increased over normal and the
difference between specific activities of vascular and extravascular albumin would be expected
to be much greater than in normals. From previous determinations of slopes and intercepts of
the plasma curve (Katz et al., 1963), the calculated EVjV specific activity ratios ranged from
3 to 4 for rats with mild proteinuria to as high as 8 when proteinuria was severe. We found,
however (Sellers et al., 1968), that 3 days after radioactive iodine-labelled albumin injection the
specific activity of total extravascular albumin became equal to that in plasma. The specific
activity of ascitic fluid behaved differently, attaining values up to eight times that in plasma.
The amount of albumin in this fraction is too small significantly to affect the average
value.
Results with another group of rats (experiment 2 of Table 9) in which proteinuria was severe
and the terminal half-life of plasma albumin was 14 h (as compared to 50-60 h in normal rats
and about 25 h in the group of nephrotic animals previously studied) is shown in Fig. 3.
18·0
14·0
10·0
Z.
:~
<:;
c
6·0
0
:t
0
Q)
a.
'"
4·0
[) Carcass
n Skin
.~
(;
Q;
a:
3'0
I Ascitic fluid
2·0
',0
nD
on nO ~
o
42
54
66
Time (h)
FIG. 3. Relative specific activity of extravascular albumin in carcass, skin and ascitic fluid of
severely nephrotic rats. Two rats were killed at each period, and the bars indicate the range
observed in the two values.
A group of sixteen rats were killed at intervals up to 96 h after injection of radioactive iodinelabelled albumin and albumin content and specific activity in blood, carcass, skin and ascitic
fluid measured. Specific activity of tissue albumin became nearly equal to that in plasma within 2-3 days, but in ascitic fluid it was 10-18 times that in plasma and remained so for the
duration of the experiment. The skin of these animals was oedematous, oozing fluid, but the
specific activity of skin albumin resembled that in muscle and plasma.
Due to the marked variability in the clinical state of the severely nephrotic rats, accurate
estimates of slopes and intercepts of the plasma curve are difficult and kinetic calculations
Extravascular albumin in the rat
721
subject to considerable error. Using a conservative estimate it is calculated that total extravascular albumin should be one-half that in the vascular pool and its specific activity at least
6·5 times that in plasma. These animals were not in metabolic steady state and the plasma
albumin concentration was decreased by up to 30% in 3 days, averaging a loss of about 0'3%
per h. However, the replacement rate of their plasma pool was some 20% per h. Thus while
calculations based on steady state assumptions must be used with caution, deviation from
steady state could not account for the very large discrepancy between observed and calculated
values.
DISCUSSION
Extravascular albumin mass
There are many studies in which estimates of extravascular albumin were attempted. The
literature is listed in the reviews of Rothschild, Oratz & Schreiber (1969), Wetterfors (1965a),
and the text of Schultze & Heremans (1966). The investigations were often incomplete and the
results conflicting. Practically all studies estimated extravascular albumin from single or double
isotope assay, frequently without attention to the specific activity of tissue albumin. Ours is,
to our knowledge, the only attempt to quantitate tissue albumin by immunochemical methods.
The most thorough study of extravascular albumin appears to be that of Oeff & Koenig
(1956) in rats. Two days after the injection of homologous [1 3 1I]albumin the animals were
exsanguinated and thoroughly perfused to remove residual blood. Radioactivity in a large
number of organs (but not skin) was measured, and the reported recovery of the injected activity
was quantitative. About 50% was excreted in urine in 48 h, 11% was in blood, 33% in the
cadavar, and less than 2% in the viscera. Thus the EVjV ratio is 3. This is a minimal value,
since at 48 h the specific activity of skin and especially gut is much less than that in blood (see
Fig. 2). The albumin content of muscle is reported to be 3·1 mgjg, a value in reasonable agreement with our own.
Wetterfors (1965b), using a double isotope procedure, determined the albumin content of
several tissues of dogs 5-8 days after injection. The muscles contained 3-4 mgjg and the
gastrointestinal tract 7-10 mgjg, values very similar to our findings in rats. Skin was not
assayed. Using a value of 4 mgjg for an average albumin content in blood-free dog tissues, we
estimate from his data that the l l-kg dogs contained some 22 g circulating albumin and at
least 45 g in the tissues, and the EVjV ratio is 2 or more. Wetterfors calculated that the extravascular albumin is 1·3 times that in blood. His calculation (Table 1 of his paper) is of interest
and requires comment. The ll-kg dogs contained about 900 ml of blood (500 ml plasma), and
total plasma albumin (43 mgjml) was 22 g. Tissues sampled and analysed by the author constituted nearly half of the blood-free body and were found to contain about 20 g of albumin.
Skin was not analysed but arbitrarily was assumed to contain 30% of the plasma content, or
6·6 g. The choice of this figure is obscure. The reference is to the work of Rothschild, Bauman,
Yalow & Berson (1955) in humans, who found the 'concentration' of total extravascular
albumin in skin to be 30% that in plasma, or for the dogs, 12 mgjg. If skin makes up 10% of
body weight, it contains about 13 g. Including blood and skin (10%), only 70% of body weight
was accounted for. The remaining 30% are, of course, not void of albumin and cannot be
neglected.
It appears that total body albumin content in rats and in dogs is similar, 5-6 gjkg, but the
722
J. Katz et al.
concentration of plasma albumin in rats is less than in most other species. Albumin concentration in rat plasma listed in a handbook compilation (Altman & Dittmer, 1961) is similar to ours
(29-35 mgjml), but higher values have been reported by several workers. Thus Oeff & Koenig
(1956) report 45 mgjml. The high EVjV ratio in rats reflects their low plasma levels, 1·3 gjkg
as compared to 1'8-2 gjkg in dogs.
The only attempt to measure albumin in interstitial fluid appears to be that of Creese,
D'Silva & Shaw (1962). They obtained a few microlitres of interfibre fluid by micropuncture
from muscles of guinea-pigs and estimate the albumin content to be half that in plasma. We
find (Table 7) the interstitial fluid albumin concentration in extracellular water of muscle and
skin to be about 55% that in plasma. The albumin content of cervical and thoracic duct lymph
is stated (Schultze & Heremans, 1966) to be half that of plasma. It is likely that this represents
the albumin concentration of the tissue fluids drained by this lymph.
Evaluation of extravascular albumin by isotopic methods
The most commonly used method for the evaluation of albumin mass and exchange kinetics
is by calculation from the slopes and intercepts of the plasma curve. The calculation is based
on the assumption of a multicompartmental model, discussed in detail elsewhere (Schultze &
Heremans, 1966; Nosslin, 1964; Katz et al., 1970).
Our work, however, reveals large discrepancies between direct measurement and calculated
values. In normal rats the estimates of the EVjV ratios differ by a factor of 3, and in nephrotic
rats with severe albuminuria, by a factor of 10 or more. Also in rats mathematical analysis
indicates usually one major extravascular pool. By direct sampling we have found two major
compartments of roughly equal size (carcass and skin) with greatly different exchange kinetics,
and a smaller pool with a slower rate of exchange (gut). Mathematical analysis requires equal
specific activity of extravascular and vascular albumin at a time when total radioactivity in the
extravascular pool is maximal. By direct measurement we find that at this maximum, the
average extravascular specific activity is no more than 70% of that in the circulation. The
actual rate of change of specific activities in the three major extravascular compartments bears
little resemblance to the average composite curve. Also, according to theory, the specific activity
of extravascular albumin should attain at distribution equilibrium a value higher than the
specific activity of plasma albumin and the ratio should be increased in conditions where
catabolic rates are elevated. In normal rats the theoretical ratio is about 1,5, increasing in
severely nephrotic animals to well over 5. We find the ratio of specific activities under all conditions to be 1·0. Our findings make the conventional model of plasma albumin metabolism
untenable, at least in the rat.
Of special interest is the high sustained specific activity of ascitic fluid. The amount of
albumin in this fraction is small, but its exchange kinetics are startlingly different from those in
the rest of the body. The albumin of ascitic fluid is readily sampled, and its exchange kinetics
have been studied extensively in humans, mainly in hepatic cirrhosis (Dykes, 1968; see also
Schultze & Heremans, 1966). The specific activity was found to equal that in plasma and only
rarely somewhat to exceed it. However, Gitlin, Janeway & Farr (1956) sampled oedema fluid
of nephrotic children and reported specific activities 8-30 times that in circulation. Such
values, to our knowledge, have not been reported elsewhere. Gitlin et al. took this high specific
activity to represent that of total extravascular albumin. Our findings in the rat indicate that
this assumption may be incorrect.
Extravascular albumin in the rat
723
Alternate models ofplasma albumin metabolism
The multicompartmental model is the simplest representation of albumin metabolism
consistent with the exponential plasma curve. The physiological validity of this model was
questioned by Reeve & Bailey (1962), who used a very different model of plasma albumin
metabolism and a different mathematical analysis. They show that their model is fully compatible with the exponential plasma curve. It also predicts higher specific activities of extravascular than vascular albumin, which is not in accord with our data.
We have explored various models with the aid of a digital computer. Very good fits to the
specificactivity curves of Fig. 2 and to the urinary excretion curve were obtained with a model
in which newly formed albumin enters each pool directly at rates approximately proportional
to the mass of albumin in the pool. It is difficult, however, to conceive how newly synthesized
albumin from the liver would bypass the circulation in its entry into the extravascular space
of skin and muscle.
If our findings are confirmed, a major revision of current concepts of plasma protein metabolism will be required. Our results stress the importance of direct study of extravascular plasma
proteins by direct sampling techniques.
ACKNOWLEDG MENTS
This work was supported by Research Grant AM-07633-06 and General Research Support
Grant RR-05468, both from the United States Public Health Service.
REFERENCES
ALTMAN, P.L. & DITTMER, D.S. (Eds) (1961) Blood and Other Body Fluids. Federation of the American Society
of Experimental Biologists, Washington, D.C.
BERSON, S.A. & YALOW, R.S. (1957) Distribution and metabolism OfI 1 3 1 labeled proteins in man. Federation
Proceedings, 16, suppl. 1, 13-18.
CREESE, R., D'SILVA, J.L. & SHAW, D.M. (1962) Interfibre fluid from guinea-pig muscle. Journal of Physiology,
162,44-53.
DEBRO, J.R., TANNER, H. & KORNER, A. (1957) The determination of serum albumin and globulin by a new
method. Journal ofLaboratory and Clinical Medicine, 50, 728-732.
DYKES, P.W. (1968) The rates of distribution and catabolism of albumin in normal subjects and in patients with
cirrhosis of the liver. Clinical Science, 34, 161-183.
GITLrN, D., JANEWAY, C.A. & FARR, L.E. (1956) Studies on the metabolism of plasma proteins in the nephrotic
syndrome. Journal of Clinical Investigation, 35, 44-55.
KATZ, J. & BONORRIS, G. (1968) Electrolytic iodination of proteins with 1'25 and 1'31. Journal ofLaboratory and
Clinical Medicine, 72, 966-970.
KATZ, J., SELLERS, A.L. & BONORRIS, G. (1963) Albumin metabolism in nephrotic rats. Journal ofLaboratory
and Clinical Medicine, 62, 910-934.
KATZ, J., SELLERS, A.L. & BONORRIS, G. (1964) Plasma albumin metabolism during transient renin proteinuria.
Journal ofLaboratory and Clinical Medicine, 64,709-716.
KATZ, J., SELLERS, A.L., BONORRIS, G. & GOLDEN, S. (1970) Studies on the extravascular albumin of rats. In:
Plasma Protein Metabolism: Regulation of Synthesis, Distribution and Degradation (Ed. by M. A. Rothschild). Academic Press, New York and London.
KINGSLEY, G.R. (1942) The direct biuret method for the determination of serum protein as applied to photoelectric and visual colorimetry. Journal ofLaboratory and Clinical Medicine, 27, 840-845.
LOWRY, O.H., ROSEBROUGH, N.J., FARR, A.L. & RANDALL, R.J. (1951) Protein measurement with the Folin
phenol reagent. Journal of Biological Chemistry, 193, 265-275.
c
724
J. Katz et al.
MANCINI, G., CARBONARA, A.O. & HEREMANS, J.F. (1965) Inununochemical quantitation of antigens by simple
radial immunodiffusion. International Journal of Immunochemistry, 2, 235-254.
MUNRO, H.N. & DOWNIE, E.D. (1964) Extraction of protein from tissues during treatment with organic lipid
solvents. Archives of Biochemistry and Biophysics, 106, 516-524.
NOSSLIN, B. (1964) In: Appendix to Metabolism ofHuman Gamma Globulin (Ed. by 8. B. Andersen). Blackwell
Scientific Publications, Oxford.
OEFF, K. & KOENIG, A. (1956) Untersuchungen mit radioaktiven Serumeiweissfractionen: Locale capillarpermeabilitet und austauschbares Albumin. Zeitschrift fur die gesamte experimentelle Medizin, 127,484-498.
PETERS, T. (1958) The isolation of serum albumin from specific precipitates of serum albumin and its rabbit
antibodies. Journal of the American Chemical Society, 80, 2701-2702.
REEVE, E.B. & BAILEY, H.R. (1962) Mathematical models describing the distribution of P31- albumin in man.
Journal ofLaboratory and Clinical Medicine, 60, 923-943.
ROTHSCHILD, M.A., BAUMAN, A., YALOW, R.S. & BERSON, S.A. (1955) Tissue distribution OfI 13 1 labelled serum
albumin following intravenous administration. Journal of Clinical Investigation, 34, 1354-1358.
ROTHSCHILD, M.A., ORATZ, M. & SCHREmER, 8.S. (1969) SeJ:1lIIl albumin. American Journal ofDigestive Diseases,
14,711-744.
SCHULTZE, H.E. & HEREMANS, J.F. (1966) Molecular Biology ofHuman Proteins. Elsevier, Amsterdam, London
and New York.
SELLERS, A.L., KATZ, J., BONORRIS, G. & OKUYAMA, S. (1966) Determination of extravascular albumin in the
rat. Journal ofLaboratory and Clinical Medicine, 68, 177-185.
SELLERS, A.L., KATZ,J. & BONORRIS, G. (1968) Albumin distribution in the nephrotic rat. Journal ofLaboratory
and Clinical Medicine, 71, 511-516.
SWAN, R.C., MODISSO, H. & PITTS, R.F. (1954) Measurement of extracellular fluid volume in nephrectomized
dogs. Journal of Clinical Investigation, 33,1447-1456.
WETTERFORS, J. (1965a) Albumin. Acta Medica Scandinavica, 177, suppI. 430.
WETTERFORS, J. (1965b) Catabolism and distribution of serum-albumin in the dog. Acta Medica Scandinavica,
177, 243-256.