Thirst and solute excretion: their effectiveness
in osmostatic control of body fluid
EWA SZCZEPANSKA-SADOWSKA,
WIKTOR
NIEWIADOMSKI,
JADWIGA
SOBOCINSKA,
AND STANISLAW
KOZLOWSKI
Departments of Applied Physiology, Institute of Physiological Sciences, School of Medicine;
and of Medical Research Center, Polish Academy of Sciences, Warsaw 00-730, Poland
SZCZEPA&KA-SADOWSKA,
JADWIGA
SOBOCI~~SKA,
AND
EWA,
WIKTOR
NIEWIADOMSKI,
STANIS~AW
KOZ~;OWSKI.
Thirst
and solute excretion: their effectiveness in osmostatic control
of body fluid. Am. J. Physiol. 244 (Regulatory
Integrative
Comp. Physiol. 13): R23-R30,1983.-The
effect of intravenous
infusion of hypertonic NaCl at three different rates (series I, 4.0
mmol/min;
series II, 8.0 mmol/min;
series III, 16.0 mmol/min)
on osmotic thirst threshold
and postloading
restitution
of
plasma osmolality (PO& has been determined
in dogs. Osmotic
thirst threshold increased proportionally
to the rate of delivery
of the osmotic load. Relative suppression of osmotic thirst at
the higher rates of infusion was temporary
and largely disappeared within 10 min after the end of hypertonic
infusion.
During the postloading
period excretion of osmotic load and
water injake was proportional
to the magnitude
of the load
administered.
However, the animals with high osmotic thirst
threshold stopped drinking at a higher P,,, than those with a
low osmotic thirst threshold. In series I and II, P,, decreased
during I h to a level not significantly
different from the preloading value. In each series renal excretion of osmotic load was
more important than water intake for restitution
of P,,, during
1 h. Amount
of water ingested during this period was significantly smaller than that theoretically
required to restore resting
P,,, when excretion of osmotic load was not taken into account.
However, total amount of water intake and osmotic load excreted during 1 h were together well adjusted to restore preloading P,,,.
osmotic thirst;
osmoregulation;
hypertonicity
DEHYDRATION
of the central nervous system
is one of the primary causes of thirst (see Ref. 4). The
first quantitative
osmometric analysis of thirst was proposed by Wolf (15). This analysis was based on the
assumption that cells (implicit: central nervous system
cells) behave like perfect osmometers, i.e., there is an
instantaneous
exchange of water between extra- and
intracellular compartments.
The osmometric equations
elaborated by Wolf (15) render possible calculation of an
osmotic thirst threshold. The latter is expressed as a
percentage decrement in the cellular water content,
which elicits the urge to drink water during an intravenous infusion of hypertonic saline. However, Wolfs
analysis is lacking data proving that the magnitude of
the osmotic thirst threshold is not influenced by a change
in the rate of administration
of the osmotic load. Recent
evidence suggesting the possibility of time-dependent
cellular adaptation to hypo- and hyperosmolality
(2, 5)
CELLULAR
0363-6119/83/0000-0000$01.50
Copyright@
1983 the American
Physiolc
as well as th .e finding that brain water permeability may
be modified by various factors (8) gives rise to the question w rhether the magnitude of the thirst thresh01 .d depends on the rate of developme nt of hype rtonicity . The
present study was designed to elucidate this issue. The
second purpose was to investigate the dynamics of restitution of resting osmolality of body fluid after its rapid
elevation. The quantitative
relationship
between increase in plasma osmolality, rate of drinking, volume of
water ingested, and rate of excretion of the osmotic load
has never been a subject of pertinent analysis. Finally,
we attempted to find out whether resting levels of plasma
osmolality and its spontaneous variations are rel .ated to
the magnitude of the experimentally found osmotic thirst
threshold.
METHODS
The following abbreviations are used in this study:
P osm9 plasma osmolality (mmol/l) ; Posmi, initial plasma
osmolality; Posm,,plasma osmolality at the thirst threshfound increase in Posm during
old;
mosm
9 empirically
infusion of hypertonic saline; APO,,,, empirically found
increase in plasma osmolality at the thirst threshold;
calculated increase in P,sm during
AP osm,, theoretically
infusion of hypertonic saline (see Eq. 5); A’Posm,in9minimum decrement in Posm after cessation of hypertonic
infusion calculated theoretically
(see Eq. 6); A’Posm.,,,,
maximum decrement in P osmafter cessation of hypertonic
infusion calculated theoretically (see Eq. 7) aP,,m,, increment in Posmfound at the end of experiment; Vu, urine
excretion (liters); L i-&o,water load calculated by subtracting volume of urine excreted during hypertonic infusion
from volume of the fluid infused (liters); Los,, osmotic
load calculated by subtracting number of millimoles excreted from millimoles administered during hypertonic
infusion (mmol) ; Losmt9 threshold osmotic load eliciting
drinking; Los, ex9osmotic load excreted; Los,,, osmotic
load remaining at the end of experiment; FL.., , fractional
cumulative excretion of the osmotic load; ICWi, initial
volume of intracellular water (liters); ICWf, final volume
of intracellular water; AICW, decrement in intracellular
water due to administration
of hypertonic load (see Eq.
4); TBW, total body water (liters); ECWi, initia1 volume
of extracellular water (liters); ECWf, final volume of
extracellular water; WI, water intake (liters); WI,, quantity of water actually ingested; WI,, quantity of water
Igical
Society
R23
R24
SZCZEPAT;JSKA-SADOWSKA,
theoretically required to restore body fluid isotonicity
(see Eq. 8); WI 1h, cumulative amount of water ingested
during the experiment.
NIEWIADOMSKI,
Ecw
f =
+
LH,o)
AP osmc
thirst threshold was measured in eight dogs in three basic
series of experiments in which osmotic load was delivered
by means of an intravenous infusion of 2.5, 5.0, and 10%
NaCl solution at a rate of 4.0 (series I, 24 expts), 8.0
(series II, 12 expts), and 16.0 (series III, 24 expts) mmol/
min.
Course of experiments. At the beginning, two thin
polyethylene catheters were introduced into the saphenous veins. The blood sample to measure resting extracellular osmolality (Posm)was taken. The urinary bladder
was catheterized and emptied by suprapubic pressure
and/or air flushing, and intravenous infusion of hypertonic solution of NaCl was started. Water was easily
accessible throughout
the whole experiment. The blood
samples to measure P osmwere taken and urine volume
and its osmolality measured every 10 min in the experiments of series III. The incidence and size of every draft
of water taken during the infusion was noted. The infusion was discontinued when the dog drank at least 50 ml
of water. It was assumed that at this point the osmotic
load infused was sufficient to activate the thirst system.
The blood sample was taken immediately after termination of the infusion. Three or four infusions of hypertonic
saline were run on separate days, before the experiments
proper started, to accustom the animals to the experimental situation. Sham experiments with infusion of
physiological saline were performed at random.
Osmotic thirst threshold was evaluated according to
Wolf (15) taking into account I) the time of infusion of
hypertonic saline necessary to elicit drinking response, 2)
the threshold osmotic load introduced during the infixsion, 3) the increase in plasma osmolality, and 4) the
degree of cellular dehydration
(AICW) caused by the
threshold osmotic load. The magnitude of the threshold
osmotic load eliciting drinking (Losm,) was calculated by
subtracting the milliosmoles excreted during the infusion
from the milliosmoles infused. The increment in plasma
osmolality eliciting drinking (APO,, ) was calculated by
subtracting initial plasma osmolality (Posmi)from plasma
osmolality found at the thirst threshold (Pas,), The
threshold cellular dehydration was calculated according
to Wolf (15) using the following equations
ICWi = TBW - ECWi
(0
ICWf = TBW + LH,,o- ECWf
(2)
Posm.
Posmi
+
Losm,
X
KOZj5OWSKI
+
Los-)
100
o
(4)
To assess whether there was an instantaneous
exchange of water between the intra- and extracellular
compartment
during intravenous administration
of the
osmotic load at the rates applied, theoretically expected
increments of plasma osmolality (APO,, ) were calculated
in each series for a particular time of infusion using the
following formula
Design
Measurements of osmotic thirst threshold during administration
of osmotic load at various rates. Osmotic
X
X
(ICWi - ICWf)
ICWi
AICW(%) =
The experiments were performed on 22 conscious male
mongrel dogs accustomed to the experimental situation.
Their mean body weight was 18.4 t 0.8 kg. The experiments were carried out at the same time of day with an
interval of at least 48 h. The dogs were fasted for 22 h
preceding the experiment, but they had free access to
water.
AND
(ECWi
TBW
Animals
Experimental
(TBW
SOBOCIr(JSKA,
TBW
=
X
Posm. 1 +
Losm
-
p
TBW + LH,o
osmi
(5)
The theoretically calculated increments were then compared to those found empirically and correlation coefficients between APO,, and APO,, were calculated separately for each rate of the infusion.
Postloading
Restitution
of Plasma
Osmolality
In each series the time of observation was extended up
to 1 h after the end of the infusion. During this time
blood samples to measure Posmwere taken every 10 min.
Plasma osmolality found at the thirst threshold was
taken as the reference point and the decrements from
this value (A’Posm)were calculated. Urine volume and its
osmolality were measured every 10 min. Thus, it was
possible to calculate the rate of excretion of the osmotic
load (mmol/lO min) and the osmotic load remaining in
the body. The latter value was calculated by subtracting
cumulated number of milliosmoles excreted during a
particular postinfusion period from the threshold osmotic
load. To assess the efficiency of renal excretion of the
osmotic load in restoring isotonicity of body fluids the
minimum
decrease in Posm (A’Posmmi.Jwas calculated
based on the assumption that water ingested was not
absorbed and that a decrease in Posmresulted exclusively
from renal excretion of the osmotic load. The following
formula was used
I
AP
(TBW
osm,,,
=
+
LH,o)
PosT
-
Losmex
-
TBW + LH,o - Vu
P osm t
(6)
and Vu correspond to osmotic load and
where
Losmex
volume of urine excreted after cessation of the hypertonic
infusion, respectively. To assess the efficiency of thirst
mechanism in restoring isotonicity of body fluids, the
incidence and size of every draft of water were noted and
rate of drinking as well as cumulative water intake calculated. The maximum decrement in Posm(A’Posm,,,)was
calculated based on the assumption that the total water
ingested was immediately absorbed. The following formula was used
A’Posm
max
=
(TBW
+
LH,o)
Posmt
-
Losmcx
TBW + LH,O + WI - Vu
-
P
OSmt
(7)
Calculated values of A’Posm,,, were compared to empirically found decrements in Posm(A’Posm,) .
The quantity of water theoretically required to restore
body. fluids to isotonicity (excretion of osmotic load not
OSMOSTATIC
CONTROL
OF
BODY
R25
FLUID
included) was calculated according to Wolf (15) using the
following formula
L
WI, = $= - LH,O
o*m,
CA/
’
'E 28.
I-
(8)
2L-
Since the rate of administration
of hypertonic load appeared to influence osmotic thirst threshold, 24 additional experiments were performed on 12 dogs given the
hypertonic load at a rate 8.0 mmol/min
(series IIa) to
increase the number of observations with the same rate
of administration
of L,,. Cumulative water intake, urine
and solute excretion, as well as P,,, were measured at 0,
10, 30, and 60 min after the thirst threshold.
Experimental
osmotic thirst threshold
neous variations of plasma osmolality.
RESULTS
Effect of Administration
of Osmotic Load at
Different Rates on Osmotic Thirst Threshold
The time of infusion of hypertonic saline necessary to
elicit water ingestion varied inversely to the rate of
delivery of the osmotic load (Fig. IA). However, the
threshold osmotic load eliciting drinking and the increment in P,, found at the thirst threshold were significantly lower in series I, in which the rate of administration of L,,, was the lowest, than in series II and III
(Table 1; Fig. 1B). Similarly, the decrement in cellular
EE 12.
20.
G loa0
a
16.
a-
12.
6-
a-
and sponta-
To elucidate
whether there is any relationship between the magnitude
of the experimental osmotic thirst threshold and resting
osmolality of body fluids, 15 dogs, whose osmotic thirst
threshold had been measured in a manner described in
series I, were placed in metabolic cages and fed a fixed
diet. The animals had continuous access to water. After
a period of 7 days, during which the animals were accustomed to stay in metabolism cages, blood samples were
taken for 10 consecutive days at the same time of day
just before feeding, and resting P,, was measured. Subsequently an experiment was performed, during which
food was withdrawn but water was continuously available. Spontaneous drinking was measured during the
whole experiment, and blood samples to determine P,,,
were taken every 30 min.
Measurements. TBW and ECW were calculated as a
percent of body weight. Measurements previously made
showed that under control conditions TBW and ECW
are 62.1 f 0.3 and 27.3 + 0.4% of body weight, respectively
(12). Plasma and urine osmolality were measured with a
Fiske osmometer.
Statistical analysis. Student’s t test for paired and
unpaired data and analysis of variance (single-factor
experimental design and factorial design with repeated
measures) were employed. The significance of individual
effects was determined
from a Newman-Keuls
range
statistic and Dunnett modification for multiple comparisons (14). Variation of variables over time was subdivided into trend components through the use of orthogonal polynomials (14). Regression lines were determined
by the method of least squares. Significance of difference
between regression coefficients of different experimental
treatments as well as partial correlation coefficients were
determined according to Bailey (3).
B
2 ILa
L0t
FIG. 1. Time of iv infusion
of hypertonic
saline (T) and threshold
increase
in plasma
osmolality
(AP,,,,)
necessary
to elicit drinking
in
series I (4.0 mmol/min),
II (8.0 mmol/min),
and III (16.0 mmol/min).
Empty
columns,
empiricahy
found values; dashed
columns,
calculated
values.
Means
+ SE are shown.
Probability
levels
determined
by
unpaired
t test.
TABLE 1. Osmotic load and fractional
cumulative
excretion of osmotic load at thirst threshold and
during 1 h after administration
of hypertonic load
Time
Variable
L Obln,
mm01
Series
94*
(2:)
II
(12)
III
(21)
F Losm,
%
+10
121
-120
133
+15
16+. +
(2:)
(lY,
III
(21)
After
Thirst
Threshold
+3
8
+2
4
fl
Administration
Load, min
of Hypertonic
10
20
30
40
50
60
89
*lo
113
84
It10
105
78
f10
98
73
r9
89
70
%9
80
64
%I0
64
+19
128
-t15
+19
120
+15
+I9
113
215
*18
106
214
+19
98
+14
+19
94
+14
22
+3
16
k4
10
+2
29
+4
23
+-5
17
+2
35
+4
31
+6
24
t4
40
+5
38
k8
30
+6
46
k5
46
-+8
38
+7
52
k6
53
+12
43
+7
Values are means + SE; nos. in parentheses
are no. of experiments.
fractional
cumulative
excretion
of
osmotic
load;
FL,,,
t series I vs. series III, P
* Series I vs. series III, P < 0.05;
* series I vs. series II, P c 0.05 by unpaired
Student’s
t test.
< 0.01;
L
1,;:
water eliciting drinking was significantly lower in series
I (2.03 + 0.2% of ICW) than in series III (3.3 + 0.4% of
ICW, P < 0.01). The subthreshold increase in P,,, found
at the very first draft of water was also significantly lower
in series I (3.4 + 0.7 mmol/l) than in series II (6.8 f 1.5
mmol/l, P < 0.02) and in series III (9.0 + 1.1 mmol/l, P
< 0.001, unpaired t test).
Good agreement was found between empirically determined increments in P,,, and those calculated with osmometric Eq. 5 (Figs. 1 and 2). In each series a highly
significant correlation was found between AP,,,< and
APosmc.The regression lines had similar slopes and the
regression coefficients (b) did not differ significantly.
R26
SZCZEPAf;JSKA-SADOWSKA,
Postloading
Restitution
of Plasma
Osmolality
NIEWIADOMSKI,
SOBOCIfiSKA,
AND
KOZLOWSKI
P < O.OOl]. In series I a significant
decrease in AP,,, was
observed starting from 50 min (Fig. 3, curve B, P < ‘0.05,
In each series diminution of APO,, exhibited a transient
Dunnett multiple comparisons test). In series II and III
inflexion (Fig. 3). This secondary increase in Posm was
a significant decrease in P osq was already observed at 20
observed in each experiment although it occurred at
min (P c 0.05, Dunnett multiple comparisons test). In
variable lapses of time after cessation of hypertonic inseries I and II the final decrements in Posmat the end of
fusion. In series I the secondary increase in Posm at
1 h equaled -3.5 t 0.9 and -5.6 t 1.2 mmol/l, respec20 and 30 min made A’Posm significantly smaller than
tively. These values were only nonsignificantly
lower
A’P osmmm. . Tests for trends revealed that restitution of Posm
than those necessary to equalize values of APO,, in these
in this series occurred nonlinearly and within the range
series. In series III the final decrement in PO,,, amounted
of time included in this study could be best described by
to -5.3 t 1.3 mmol/l and did not equalize APO,, , which
a third-degree equation. No significant deviation from
amounted to 10.6 t 1.2 mmol/l (P < 0.01, paired’ t test).
linearity was found in series II and III (Table 2). MultiExcretion of osmotic Load. Table 1 presents the absofactorial analysis of variance with experimental design,
lute and fractional excretion of Los, after cessation of the
subjects, and time as factors did not show significant
hypertonic infusion. Multifactorial
analysis of variance
overall variation among the three series [F(2,45) = 1.5; P
with experimental design, subjects, and time as factors
> 0.051 in empirically found decrements in PO,,. Signifidid not show significant differences among the three
cant variation was found within series [F(5,225) = 9.7;
series of experiments as related to Los, and FL osm . In each
series, Los, decreased and FL osmincreased significantly
I-SeriesI;
r=O62;
n=62;
P4LOOl
during the first 10 min after the end of the hypertonic
y= c&9x+ 2.52;
z 30infusion (P < 0.005 and P < 0.001 at 10 min and subse/'
? 28- II-Series&
r = a78; n = 31; P 4IOOl
//
quent intervals, respectively, Dunnett multiple compariy= mx+ 2.63;
sons test). Test for trends revealed that Los, decreased
-E 26II
024 E
TABLE
2. Analysis of trends components
&i 22
Q 20.
L osm
A’R,sme
wLm
I
FLosIn
Se181
I
ties
T
Trend
F
16
I
L
Q
C
II
r =
Q78; n=S2;P
C
y= 063x407;
0 ~~------,r-r-T-v-T0 2 4 6
8
10 12 14 16 18
III
20 22 24 26 28 30
4km
e
lines of calculated
(APO,,,)
on empirically
found
in plasma
osmolality
during
and at end of iv
saline in series I, II, and III (solid lines). Dotted
line (APosm, = APosm,).
2. Regression
(AP,,,J
increments
infusion
of hypertonic
line, ideal regression
Series
I
Series
0
10
20
L
Q
(mmol/l)
FIG.
L
Q
(a001
30
C
P
13.7
1.1
4.2
(LlW
<O.OOl
NS
co.05
26.9
1.1
1.1
(1,55)
~0.001
NS
NS
23.1
2.7
2.8
(1,lW
CO.001
NS
NS
F
P
296.3
0.0
0.0
(ma
co.oo1
NS
NS
126.0
0.0
0.0
(1,50)
co.oo1
NS
NS
345.8
0.0
0.0
U,lW
co.oo1
NS
NS
Trends:
L, linear;
Q, quadratic;
other abbreviations
see text. Nos.
freedom.
F
P
54.7
0.0
0.0
ww
co.oo1
NS
NS
43.3
0.0
0.0
(1,50)
co.oo1
NS
NS
94.4
0.0
0.0
(LlW
co.oo1
NS
NS
C, cubic; NS,
in parentheses
F
140.7
0.2
1.2
P
co.001
NS
NS
(1,126)
68.0
4.0
0.0
<O.ool
co.05
NS
~1,66)
203.7
19.2
1.6
co.oo1
<O.ool
NS
(1,126)
not significant.
For
represent
degree of
III
LO
50
60
FIG. 3. Postloading
decrements
in plasma osmolality
(A’P,,)
in series I and III. Curves:
A, calculated
minimum; B, empirically
found; C, calculated
maximum
decrements
in P,,, after cessation
of hypertonic
infusion.
Mean
values
t SE are shown.
Factorial
analysis
of
variance
showed
a significant
overall variation
between
[F(2,60)
= 20.8, P < O.OOl] and within
[F(5,300)
= 11.27,
P c O.OOl] variables.
Significance
levels of individual
differences
between
A, B, and C determined
from Newman-Keuls
range statistic.
AP,,mt
Af&,=
= 5.1 ,+ 0.7 (mmol/l)
-12 t
-12
rAvs
l
A,B
B
vs
P (
C
P (
005
0.01
t
10.6 2 1.2 (mmol/l)
OSMOSTATIC
CONTROL
OF
BODY
R27
FLUID
and ~~~~~increased linearly
over the time within the
range of time included in this study and could be best
described by a first-degree equation (Table 2). The equation of the fitted regression lines for L,,, were: y1 =
-0.47x1 + 91.3;yz = -0.43~2 + 122.0; y3 =
-0.57x3
+ 127.9
in series I, II, and III, respectively. The slopes of regression lines in the three series did not differ significantly.
The respective equations for FL(,~~were y1 = 0.58~1 +
16.95;yz = 0.79x2 + 6.68; y3 = 0.62& + 4.92. The slopes of
the regression lines in the three series did not differ
significantly.
The results presented in Fig. 3 show that excretion of
L,,, largely accounted for a postinfusion restitution of
P osm during the observation period. In series II (not
shown) and III the calculated minimum decrements in
P osmt did not differ significantly from those found empirically. It should also be noted that in series I A’Posm at 20
and 30 min were significantly smaller than APosme.
mm*No
significant differences among these variables were found
during other periods of time (Fig. 3).
Water intake. Volume of water ingested at the thirst
threshold did not differ significantly among the three
series (Fig. 4). During the postloading period the drinking
rate in series II and III was significantly higher than in
series I. Differences in WI between series I and III started
to be significant within 10 min after the end of hypertonic
infusion (Fig. 4). Test for trends revealed that within the
range of time included in this study cumulative water
intake in series I occurred linearly. On the other hand a
significant deviation from linearity was found in series II
and III (Table 2). The results indicate that the second.degree equations should be used in these series to approximate the relationship between cumulative water
intake and time.
Water intake did not contribute markedly to postloading restitution of Posmduring the observation period. In
each series A’P osm,was significantly smaller than A’Posm
(Fig. 3). Moreover, volume of water actually ingested
f:
-E
3
during 1 h (WI,) was in each series significantly lower
than that which should be theoretically drunk to restore
body fluid isotonicity (Table 3). On the other hand,
volume of water ingested and osmotic load excreted
during 1 h were together properly adjusted in each series
to equalize Pos,t. In series I A’PosmmaX
exceeded just significantly the value of APosmt(-7.5 t 0.5 mmol/l and 5.1
t 0.7 mmol/l, respectively, P < 0.05; paired t test). In
series II values of A’Posm and of Posmtwere -8.2 t 0.8
and 8.8 t 1.4 mmol/l, and in series III, -9.5 t 1.6 and
10.6 t 1.2 mmol/l, respectively (see Fig. 3).
Tables 4 and 5 present correlation and partial correlation coefficients between the magnitude of the osmotic
thirst threshold, water intake, and osmotic load excreted
in cumulated experiments of series II and IIa. A strong
positive correlation was found between APO,, and APO,,,
as well as between Losmt and APO,,,. A significant correlation was also found between APosmtand APO,,, found at
30 min after cessation of hypertonic infusion in experiments in which the animals stopped drinking 30 min
before the end of the experiment (Fig. 5).
max
TABLE 3. Comparison of cumulative amount of water
actually ingested during experiments and amount
of water that should theoretically be ingested
to restore body fluids isotonicity
Series
n
I
WIa
21
12
21
36
II
III
II + IIa
P
WI,
162 AI 15
221 & 32
243 t 26
195 t 15
245 t 23
(0.01
353 -+ 38
396 t 44
327 t 32
co.05
WI,
x 100/w1,
66
63
61
CO.01
<0.001
60
Values are means t SE; n, no. of experiments;
for other
see text. Student’s
paired t analysis
was employed.
abbreviations
4. Correlation and partial correlation
coefficients between variables in cumulated
experiments of series II and IIa
TABLE
280
Variables
Coefficients
n
240
200
Excluded
Y
X
N’osm,
Lo,,,
Aposm,
APosmf
L
L
L osmf
L Osmex
osmt
osmt
APosm,
WI1
h
Aposm,
WII
h
WI1
h
WI1
h
L osmt
L osmt
AICW
r
34
32
32
32
34
0.79
0.76
0.82
0.74
-0.01
Lsm,x
0.17
34
0.30
32
32
32
0.29
Lsm,x
WI1 h
P
rP
0.11
<O.ool
co.oo1
to.001
<O.ool
NS
NS
NS
NS
NS
Thirst
was stimulated
by administration
of hypertonic
load at a rate
of 8 mmol/min.
r, Correlation
coefficient;
r-p, partial
correlation;
JZ, no.
of experiments;
NS, not significant.
For other abbreviations
see text.
80
40
5. Correlation coefficients between variables
of cumulated experiments of series I, II, and III
TABLE
0
b lb io j, io io . $0
Variables
(mid
FIG. 4. Postloading
cumulative
water intake in series I, II, and III.
Mean values t, SE are shown.
Factorial
analysis
of variance
showed a
significant
overall
variation
between
[F(2,48) = 5.1; P < 0.051 and
within
[F(6,288)
= 66.9, P < 0.001) series. * Series I vs. series III, P <
0.05. A Series I vs. series II and III, P c 0.05; n Series I vs. series II and
III, P < 0.01 as determined
from Newman-Keuls
range statistic.
Aposm,
WI1
h
L osmt
AICW
P osmt
WI1
h
WI1 h
WI1
n
53
53
53
53
0.54
0.46
0.24
h
n, No. of experiments;
see text.
r
0.49
NS,
not
significant.
P
For other
<O.ool
<o.(KH
<O.ool
NS
abbreviations
R28
SZCZEPAfiSKA-SADOWSKA,
NIEWIADOMSKI,
SOBOCIr;JSKA,
AND
KOZ~O’WSKI
agreement between values of APosm found empirically
and those calculated using Q. 5, which is based on an
assumption that there exists an instantaneous exchange
of water between ECW and ICW. A highly significant
positive correlation with almost the same slopes and
intercepts of regression lines was found between measured and predicted values of APosm for each rate of
administration
of the osmotic load. Therefore, it appears
that in each series exchange of water between the ECW
and ICW compartments occurred in general according to
the characteristics
of perfect osmometers although a
possibility of some regional differences cannot be excluded. Especially, exchange of water at the level of the
central nervous system may occur in a different way. The
latter assumption may be supported by the finding of
Raichle et al. (8) that brain water permeability may be
modified by neurohormonal
factors. Finally, the rate of
cellular dehydration in series III could have been beyond
the range at which the thirst system operates under
natural conditions. Using Eqs. l-4 one can estimate that
water loss from ICW approximated in series III is 24 ml/
min, whereas an intense thermoregulatory
water loss in
the dog exposed to high ambient temperature
or subComparison of the experimental osmotic thirst threshjected to 2,4=dinitrophenol hyperthermia
varies between
old and spontaneous variations of plasma osmolality.
5 and 15 ml/min (1, 13). We speculate that with very
No significant correlation was found between mean of high rates of dehydration the reaction of the thirst system
repeated measurements of resting P,,, and magnitude of is not directly proportional to the strength of the stimuthe experimental thirst threshold in the same animals.
lus
Neither was there a significant correlation between APosmt
Results presented in Fig. 4 indicate that relative inefand variance of mean resting P,,.
ficiency of thirst observed at a high rate of development
Mean plasma osmolality in a group of 10 dogs standing
of hypertonicity is transient and largely disappears within
quietly in a Pavlov stand and having free access to water
10 min following administration
of the osmotic load.
was remarkably constant and, within a range of a 3-h Therefore this phemomenon
does not appear to be inobservation period, varied from 292.5 t 0.7 to 293.9 t 1.4 volved in long-lasting suppression of thirst under various
mmol/l. Fluctuations in P osmin individual experiments
experimental
conditions causing rapid dehydration
(6,
were more accentuated. Mean of maximum increases in 10). Nevertheless, the present results point to the imporP osm found during the observation period in individual
tance of the use of the same rate of delivery of hypertonic
dogs equaled 4.7 t 0.9 mmol/l and did not differ signifiload whenever the osmotic thirst threshold is to be
cantly from the threshold increment in Posm eliciting
compared.
drinking in the same group of animals during measurements of osmotic thirst threshold (5.2 t 0.7 mmol/l).
Postloading Restitution of Plasma Osmolality
Mean water intake in the same period of observation
Changes of plasma osmolality, water intake, and solute
was also stable and varied from 5.7 t 3.3 to 15.7 t 6.5
excretion were followed in the present,study for 1 h. This
ml/30 min.
period of observation was chosen based on previous data
from
this laboratory showing that volume of water inDISCUSSION
gested during this period equals 90% of a t,otal volume
An interesting and unexpected finding of the present
ingested in response to the threshold osmotic stimulus
study was that the magnitude
of the osmotic thirst
(11)
threshold increased proportionally
to the rate of delivery
Solute excretion and water ingestion are two principal
of the osmotic load, i.e., the lowest value of the osmotic
mechanisms involved in restitution of elevated osmolality
thirst threshold was found when the hypertonic load was of body fluids. Data presented in Table 2 reveal a linear
infused at the lowest rate.
pattern of excretion of the osmotic load during the obTheoretically,
the following possibilities should be servation period. On the other hand diminution of plasma
taken into account to explain this finding. First, a sig- osmolality exhibited in each series a transient deflection
nificant difference between APosmtin series I and III could
at 20-30 min, which in series I caused a significant
be explained by the existence of a constant latency for
deviation from linearity. It should be noted that the
drinking. However, the data presented in Fig. lA allow
observed deviation in A’Posm fit neither the pattern of
us to reject this possibility. Second, relative suppression
cumulative removal of osmotic load nor the pattern of
of osmotic thirst in series III might have been caused by cumulative intake of water. In series I a secondary inthe existence of some delay in the rate of exchange of crease in P osmwas high enough to make A’Posm signifiwater between the ECW and ICW compartments.
Howcantly even smaller than A’Posmrn*n*
. In individual experiever, the data presented in Figs. 1 and 2 show good
ments, the secondary increase in Posm sometimes ex-
OSMOSTATIC
CONTROL
OF
BODY
R29
FLUID
ceeded the magnitude of AP,,,t. A release of some osmotic material from tissues or ‘transient shift of water
may be the only explanation for this finding. Apparently,
an intrinsic shift of water or osmotic material may be an
additional factor marked1 .y influencing the course of restitution of plasma osmolality. A slowly progressing dehydration presumably causes some adaptive changes at
a cellular level accompanied by production
of idiogen
osmoles (2). Relevant to this, it should be noted that a
significant deviation from linearity in restitution of P,,,
was observed in series I, in which the osmotic load was
administered at the lowest rate.
The data presented in Tables 1 and 3 show that in
each series, i.e., independent of the magnitude of initial
deviation in P,,,, the total fraction of the threshold
osmotic load excreted during the experiments as well as
the total fraction of water actually ingested to water,
which theoretically must be drunk to restore body fluids
osmolality, were similar and approximated
50 and 60%,
respectively. Close inspection of Fig. 3 reveals that excretion of osmotic load within the range of ti .me included in
this study was relatively more important for restitution
in series I than-in series III. In series I A’P,,, .
calculated at the end of 1 h did not differ significantly
from A’Posm or from APoSmteliciting thirst in this series.
Thus, solute excretion in this series was sufficient to
abolish deviation in P,, caused by the osmotic load. In
series III A’P,,, . at the end of -1 h was significantly
smaller than AP osm+eliciting drinking.
In each series the difference between the decrease in
PO, found empirically at the end of 1 h and the corresponding A’P,,, .m;n
.... was much smaller than the difference
between A’‘Posme and A’posm
This finding suggests a
significant delay in absorpt%n of water ingested and
further confErns that excretion of osmotic load is more
important for an immediate lowering of plasma osmolality than drinking of water. On the other hand, within a
longer period of time water ingestion may account for
the larger part of total postloading lowering of plasma
osmolality. Comparison of values of APosmmax
and threshold increments in P osmnecessary to induce drinking disclose an interesting finding that, in spite of a long delay
in reabsorption of water, its amount taken during 1 h is
(in conjunction with excretion of the osmotic load) precisely adjusted to equalize initial deviation of plasma
osmolality. The mechanism coupling water ingestion and
solute excretion is at present obscure. A good adjustment
of the amount of water ingested to actual body needs
long before it is absorbed may effectively decrease body
fluids tonicity and probably argue for importance
of
gastrointestinal
control of water intake (9, 16).
In general, analysis of trend components disclosed
some differences between series with regard to water
intake and restitution of PO,,. Unfortunately,
the experimental design used in this study does not allow us to
conclude whether these differences were caused by differences in the rate of development of hypertonicity
or
in absolute magnitude of the load administered.
A strong positive correlation found between threshold
osmotic load and osmotic load excreted suggests that
excretion of osmotic material was proportional
to the
administered
load. However, the relationship
between
these two parameters was not strong enough to reduce
of posm
mm
mm
L osmc to comparable values in animals with a different
mag’nitude of the osmotic thirst threshold, and correlation between L osm, and Los,, was also strongly positive.
Surprisingly enough, in animals subjected to infusion
of hypertonic saline-at the same rate, total cumulative
volume of water ingested did not correlate either with
Losmt or with APosm or AICW even after exclusion of the
osmotic load excreted. On the other hand, a significant
positive correlation between these parameters was found
when cumulated data of series I, II, and III were taken
into account (see Tables 4 and 5). This discrepancy can
possibly be explained when one takes into account variability of osmotic sensitivity of the thirst system in
individual animals. It may be expected that animals with
a high osmotic thirst threshold should terminate drinking
at a higher plasma osmolality than those with a lower
osmotic thirst threshold. A strong positive correlation
between APosmt or Los,t on the one hand and APO,,, on
the other hand may support this assumption. A highly
significant correlation was also found between APosmtand
an increment in Posmfound at 30 min after cessation of
the infusion in experiments in which animals stopped
drinking 30 min before the end of the experiments. The
latter finding indicates that animals with a high osmotic
thirst threshold stop drinking at a relatively high increment in Posmeven when they have free access to water.
In the case of cumulated data of series I, II, and III the
same animals were subjected to increasing osmotic load,
and the relation between APosmtand WI could have been
more visible. Dependence of total intake of water on
individual osmotic reactivity of thirst could probably
explain individual variation in the drinking response to
the same amount of hypertonic
NaCl observed by
Holmes and Gregersen (7).
Inspection of spontaneous changes of plasma osmolality and water intake in animals having free access to
water indicates that, in spite of big fluctuations in individual experiments, mean plasma osmolality and water
intake remained remarkably constant. It is worth noting
that the mean of maximum increments in Posmin individual experiments did not differ significantly from the
magnitude of the osmotic thirst threshold (APosmt)found
in the same animals. Thus, osmotic reactivity of the
thirst system probably to some extent determines increase of spontaneous P osmeliciting water intake. On the
other hand, increasing the osmotic thirst threshold does
not seem to play a significant role in long-term control of
a level at which plasma osmolality is maintained, since
there was no correlation either between APos,t and the
resting level of plasma osmolality or between APosmtand
the range of variability of day-to-day measurements of
resting PO,,. Accordingly, animals with a high experimental osmotic thirst threshold are able to maintain resting
Posmat a comparable level to those with a low threshold.
It appears that an appropriate renal excretion of water
and solutes may markedly contribute to a long-term
of
regulation of resting plasma osmolality. Stimulation
thirst by extracellular dehydration shoul .d also be taken
into account.
In summary, the present results indicate that osmotic
sensitivity of thirst may be temporarily suppressed at a
high rate of development of hypertonicity.
We also find
that renal excretion of the osmotic load plays a more
R30
SZCZEPAT;JSKA-SADOWSKA,
important role than water intake in restitution of plasma
osmolality during the 1st h after administration
of the
hypertonic load. Excretion of osmotic load and water
ingested are together precisely adjusted to restore elevated osmolality of body fluids to resting level. Although
magnitude of the osmotic thirst threshold in individual
animals appears to determine deviation of plasma osmolality tolerated by the thirst system and contribution
of thirst to restitution
of resting plasma osmolality, it
NIEWIADOMSKI,
SOBOCINSKA,
AND
KOZLOWSKI
does not seem to play a significant role in long-term
control of a level at which plasma osmolality is controlled.
We gratefully
acknowledge
the
Brzek.
This study was partly
financially
RMZ-/no
1.83 and 10.4 PAN.
Received
21 December
1981; accepted
technical
supported
in final
assistance
within
form
of Kazimiera
the projects
30 April
lo-
1982.
REFERENCES
1. ADOLPH,
E. F. Signs and symptoms
of desert
dehydration.
In:
Physiology
of Man
in the Desert,
edited
by E. F. Adolph.
New
York: Interscience,
1947, p. 226-240.
2. ARIEFF,
A. J., R. GUISADO,
AND V. C. LAZAROWITZ.
Pathophysiology of hyperosmolar
states. In: Disturbances
in Body FZuid Osmolality,
edited
by T. E. Andreoli,
J. J. Grantham,
and F. C.
Rector.
Bethesda,
MD: Am. Physiol.
Sot., 1977, p. 217-225.
3. BAILEY, N. T. J. StatisticaL
Methods
in Biology.
London:
English
Universities,
1961, p. 91-99 and 136-149.
4. FITZSIMONS,
J. T. The PhysioZogy
of Thirst
and Sodium Appetite.
Cambridge:
Cambridge
Univ.,
1979, p. 128-190.
5. GRANTHAM,
J. J. Pathophysiology
of hyposmolar
conditions:
a
cellular
perspective.
In: Disturbances
in Body Fluid
OsmoZaZity,
edited
by T. E. Andreoli,
J. J. Grantham,
and F. C. Rector.
Bethesda,
MD: Am. Physiol.
Sot., 1977, p. 217-225.
6. GREENLEAF,
J. E., AND F. SARGENT II. Voluntary
dehydration.
J.
AppZ. Physiol.
20: 719-724,
1965.
7. HOLMES,
J. H., AND M. J. GREGERSEN.
Observations
of drinking
induced
by hypertonic
solutions.
Am. J. PhysioZ.
162: 326-337,195O.
8. RAICHLE,
E. M., R. L. GRUBB,
AND J. 0. EICHLING.
Central
neuroendocrine
regulation
of brain water permeability.
In: CerebraZ
VascuZar
Smooth Muscle
and its ControZ.
Ciba Foundation
Symposium
56. Amsterdam:
Elsevier/North-Holland,
1978, p. 219-235.
9. ROLLS, B. J., R. J. WOOD, E. T. ROLLS, H. LIND, W. LIND, AND J.
10.
11.
12.
13.
14.
15.
16.
G. G. LEDINGHAM.
Thirst
following
water deprivation
in humans.
Am. J. Physiol.
239 (Regulatory
Integrative
Comp. Physiol.
8):
R476-R482,
1980.
ROTHSTEIN,
A., E. F. ADOLPH, AND J. H. WILLS. Voluntary
dehydration.
In: Physiology
of Man
in the Desert,
edited
by E. F.
Adolph.
New York: Interscience,
1947, p. 254-270.
SOBOCI~SKA,
J. Gastric
distension
and thirst:
relevance
to the
osmotic
thirst
threshold
and metering
of water intake.
Physiol.
Behav.
20: 497-501,
1978.
SOBOCI~~SKA, J., AND E. SZCZEPA~~SKA-SADOWSKA.
Normal
values
of body fluid volume, blood constituents,
hemodynamic
parameters,
osmotic
reactivity
of the antidiuretic
system and the thirst
mechanism in unanaesthetized
dogs. Acta
Physiol.
PoZ. 26: 613-624,
1975.
SZCZEPA~~SKA~~ADOWSKA,
E. Osmotic
thirst suppression
during 2,4dinitrophenol
(DNP)
hyperthermia
in the dog: Pfluegers
Arch. 355:
165-174,
1975.
WINER,
B. J. StatisticaL
Principles
in Experimental
Design.
New
York: McGraw,
1970, p. 46-104,
105-139, and 298-378.
WOLF, A. V. Osmometric
analysis
of thirst in man and dog. Am. J.
Physiol.
161: 75-86, 1950.
WOOD, J., S. MADISON,
E. T. ROLLS, B. ROLLS, AND J.. GIBBS.
Drinking
in rhesus monkeys:
roles of presystemic
factors in control
of drinking.
J. Comp. Physiol.
Psychol.
94: 365-374,
1980.