Biochem. J. (1971) 121, 293-297
Printed in Great Britain
293
The Diffusion Capsule, a Novel Device for the Addition of
Constant Rate to a Liquid Medium
a
Solute at
a
ITS APPLICATION TO METABOLIC REGULATION
By S. J. PIRT
Department of Microbiology, Queen Elizabeth College, University of London, Campden Hill,
London W.8, U.K.
(Received 20 August 1970)
The diffusion capsule consists of a cylindrical container that can be completely
filled with a solution and sealed with a small semi-permeable membrane at one end.
In use, the capsule is immersed in an agitated liquid. Experiments on concentrated
solutions in the capsule showed that, contrary to diffusion theory, the rate of
diffusion of solute (sugars or amino acids) out of the oapsule remained virtually
constant until about 65% of the solute had diffused out of the capsule. Thus the
device has been used to maintain constant material feed rates for periods exceeding
30h. The capsule is a simple and compact substitute for a pump and is superior to
a pump for small feed rates in many applications. The capsule greatly extends the
scope of the shake-flask culture technique for micro-organisms in that substratelimited growth, possibly the aspect of greatest interest, is readily achieved simply
by dropping in the flask a capsule containing the substrate. Diffusion feed should
facilitate study of the metabolism of toxic substrates: also it is likely to provide an
improved means for supplying a pulse of tracer to a culture.
Continuous-feed devices are basic tools of
physiological science and a great variety of such
devices are in use; nevertheless for certain important applications no suitable continuous-feed device
is available. All the methods in general use are
based on some form of pump. Pumps, however, are
too cumbersome and costly for multiple cultures of
micro-organisms on the laboratory flask scale and,
moreover, at the lower feed rates required the
discontinuity of mechanical pumps becomes unacceptable. As a solution to this problem the
'diffusion-feed' technique was developed. In the
method the material is fed into a medium from a
submerged capsule. The material diffuses out of
the capsule through a small semi-permeable membrane and, for a long period, the diffusion rate is
regulated to a constant value by what appears to
be a novel osmotic phenomenon. Earlier methods
of substrate feeding by dialysis have depended on
the use of special flasks divided into compartments
by dialysis membranes or simply by immersion of
dialysis tubing in the medium. In all of these
methods the normal simple diffusion kinetics have
been found to apply (Schultz & Gerhardt, 1969) so
that a constant feed-rate is possible only when the
concentration difference across the membrane is
maintained constant; this limitation is overcome
in the diffusion capsule.
DESIGN OF DIFFUSION CAPSULE
The capsule is illustrated in Fig. 1. It consists of a
nylon cylinder with an orifice at one end tightly sealed
with a membrane held between an '0' ring and a cap. The
membrane was cut from Visking cellulose dialysis tubing.
It had a thickness of 0.05mm when dry and swelled about
50% on wetting. The other end of the capsule had a fullbore orifice to facilitate cleaning and filling. The seals on
the caps must be capable of withstanding an internal
pressure of 15001b/in2, which can be generated by the
osmotic effect.
RESULTS AND DISCUSSION
Diffusion rate8 of glucose and ly8ine. It was
originally expected that the rate of diffusion of a
solute out of the capsule would not be constant but
that it would decrease in proportion to the difference
between the internal and external solute concentrations as demanded by simple-diffusion theory.
Thus if s = amount of solute in the capsule at time
t; el = concentration of solute in capsule; ce = concentration of solute in external medium; A =
surface area of membrane, then according to simple
diffusion the flux of solute out of the capsule should
S. J. PIRT
294
I
I
membrane
e-,
. ~~~
1971
T
<
rI
.g
'O.' ring
0
6
0bo
._0
Cs
Neoprene
gasket
10
0
S
0
0
Scale 2
V)
.
10 mm
Fig. 1. Cross-section through a diffusion capsule. The
membrane orifice diameter is 4.0mm; the capacity of the
capsule, 1.2ml. The body is made from nylon rod. The
'O' ring [jj (outer diam.) x ^ (inner diam.) x & in] is fixed
in its groove with Araldite. Each cap is screwed on a
thread about 12mm long (22 threads/in); the number of
threads/in is not critical. For the 7.5mm orifice the 'O'
ring was i (outer diam.) x i (inner diam.) x * in.
0D
x
1
,-'
f
4
5 6 '
24
Time (h)
Fig. 2. Glucose diffusion from capsule with membrane
orifice diameter 7.5mm. One membrane layer; external
water volume, 100ml; temperature, 37°C. Initial glucose
concentration in capsule (%, w/v): 0, 3.13; x, 6.25;
El, 12.5; 0, 50.
0
2
3
be given by -ds/dt = PA(ol - c,) where P is the
permeability coefficient of the membrane. It was
found, however, that instead of falling progressively
the rate of diffusion remained nearly constant until
the internal solute concentration fell by some 65%;
indeed, initially the diffusion rate could even increase for a time. The experimental evidence for
ba
this novel behaviour is outlined below.
To determine the diffusion rate of glucose, the
filled capsule was dropped into a conical flask
.S0O
containing water and agitated on a rotary incubated
shaker. The glucose in the external medium was
.6
x
determined in 1 ml samples by the glucose oxidase
00
method. The rates of diffusion of glucose are depic4
0
ted in Figs. 2 and 3. With a membrane orifice of
v
7.5mm diameter (Fig. 2) the rate ofglucose diffusion
remained virtually constant for 5-6h, in which
0
time 60% or more of the glucose diffused out of the
30
0
2 4 6 8 10 12 -4i25
capsule. The duration of the constant diffusion rate
Time (h)
was prolonged to 12h or more by decreasing the
membrane orifice diameter to 4mm (Fig. 3). With Fig. 3. Glucose diffusion from capsule with membrane
diameter 4mm. One membrane layer; external
the higher glucose concentrations an acceleration orifice
water volume, 100ml; temperature, 370C. Initial glucose
in the diffusion rate occurred initially. This was concentration
in capsule (%, w/v): x, 30; o, 50.
more marked with the smaller (4mm diam.) orifice
-
(Fig. 3).
The influence of the initial glucose concentration
on the maximum diff-usion rate is shown in Fig. 4.
According to the nominal area of the membrane,
decreasing the orifice should decrease the diffusion
rate by the factor 3.5, wherease the actual factor
was about 2. One reason for this discrepancy is that
the membrane area was somewhat indeterminate
DIFFUSION FEED
Vol. 121
because of the difference in diameter of the 'O' ring
seal and the orifice. Secondly, there was evidence
that on decreasing the size of the membrane
orifice the deviation from simple diffusion was
accentuated. When the orifice diameter was
decreased to 2.5mm the diffusion rate of glucose
(50%, w/v) was found to fluctuate to a marked
degree over a period of about 4h. It was therefore
concluded that for a constant feed-rate there is a
critical minimum diameter for the orifice between
2.5 and 4mm.
100 r
0
S
0bo
._
.,.
0
0
S
860
6bj(
*RcS
0
10
20
30
40
50
Initial conen. of glucose in capsule (%, w/v)
Fig. 4. Maximum glucose-diffusion rate as a function of
initial glucose concentration in the capsule. Temperature,
37°C; o, 7.5mm-diam. membrane; x, 4mm-diam. membrane.
295
The diffusion of lysine out of the capsule was
followed by the quantitative ninhydrin method
(Yemm & Cocking, 1955). The results are depicted
in Fig. 5. The duration of the constant feed-rate was
considerably longer than was the case with glucose
under similar conditions. On increasing the number
of layers of membrane from one to three the diffusion
rate varied roughly in inverse proportion to the
number of layers.
No satisfactory explanation of the deviation from
simple diffusion kinetics can be offered. The main
anomaly to be explained is the apparent increase in
the membrane permeability with time, which
compensates for the fall in the concentration
difference. The diffusion capsule differs from
previous dialysis feed devices in that large osmotic
pressure differences may be generated across the
membrane. But it is difficult to see how this can be
the cause of the anomaly because it was just as
marked with a glucose concentration of 5% (w/v)
as with 50% (w/v). Also, if deformation of the
membrane by the pressure difference were the cause
of the anomaly, this should be greater the larger the
orifice, since the same material covering the smaller
orifice should be less deformed. However, the
anomalous behaviour was just as marked with the
smaller orifice.
Application2 of the diffu?son cap8ule. The capsule
is the first convenient means for obtaiiing the low
rates of substrate feed required for multiple cultures
of microbes and cells on the shake-flask scale. This
opens up many new possibilities for controlled
experiment, particularly in the physiological field,
where regulation of substrate supply is fundamental
to the control of metabolism and elucidation of its
mechanism.
An example of the use of the capsule for control
of the glucose metabolic rate in E8cherichia coli is
shown in Figs. 6(a) and 6(b). Minimal media
a
0
0-_
S
0
0
0
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
Time (h)
Fig. 5. Diffusion of lysine monohydrochloride from capsule with membrane diameter 7.5mm. External water
volume, lOOml; temperature, 25BC. Initial lysine monohydrochloride concentration in capsule, 5% (w/v).
Number of membrane layers: *, 1; x, 2; O, 3.
S. J. PIRT
296
Pa
2.8
2.8
2.4
2.4
2.0
P-z 2.0
1.6
1.6
0 1.2
1971
(b)
c;
4
0 1.2
0.8
0.8
0.4
0.4
0
1
0
0
2
I
1
3
2i
2
4
3
3
5
4
4
0
6 (forA)
o
f
5 6 (for B)
2
4
6
0
2
4
1
2
3
4
0
1
2
3
5 6 (for.C)
4
5
6 (fo r
D)
10
I2 24
10
6
8
4I
6
0 24
6
8
~10
4
6
8
0I
0
8
0
2
36 48 (for E)
12 24
1
128
36 48(for F)
12 24
10
36 48 (for G)
12 24 36 48 (for H)
Time (h)
6.
of
Fig. Regulation growth rate of E. coli by diffusion feed of glucose. Opacity was used as the measure of
bacterial concentration (1 opacity unit= 0.79mg dry wt. of bacterial/ml). The cultures were shaken on a
rotary incubator shaker at 370C. The concentration of glucose in the diffusion capsules was 50% (w/v). (a) Fast
diffusion. The cultures (initial volume, 60ml) were contained in 250ml conical flasks. The capsules had
membrane orifice diameters of 7.5mm. Culture A was initially supplied with 3.34mg of glucose/ml and had no
capsule; cultures B, C and D were supplied with glucose by diffusion from capsules with one, two and three
membranes respectively. (b) Slow diffusion. Each culture had an initial volume of 165ml and was contained
in a 1 litre conical flask. The capsules had orifice diameters of 4mm. Culture E was initially supplied with
3.64mg of glucose/ml and had no capsule. Cultures F, G and H were supplied with glucose by diffusion from
capsules with one, two and three membrane layers respectively. Cultures E, G and H were duplicated; the
duplicate of culture F was lost; the mean opacities are plotted; the opacities of duplicate cultures agreed
within 0.05.
(ammonia +inorganic salts) in conical flasks were
inoculated with actively growing E. coli. Glucose
was added as the sole carbon source either initially
or by diffusion from a capsule aseptically placed in
the flask immediately after inoculation. The filled
and sealed capsules were sterilized by autoclaving
at lOlb/in2 (110OC) for 10min. For 'fast diffusion'
(Fig. 6a) a membrane of 7.5mm diameter was
used with an initial culture volume of 60m]; for
'slow diffusion' (Fig. 6b) the 4mm-diameter membrane was used and an initial culture volume of
165ml.
Curve A in Fig. 6(a) shows the normal growth
curve obtained when glucose was present initially
in the medium. Curves B, C and D show growth
curves when glucose was fed at different rates by
means of the capsules. Curves B and C, with one or
two membrane layers respectively, show constant
linear increases in growth in keeping with the
constant glucose feed-rates. The ratio of the slopes
in the linear regions of curves B and C is 1.83, which
is in fair agreement with the inverse ratio of the
membrane thicknesses. Curve D shows some stepwise increase in growth. This stepwise growth or
partially synchronized division was a feature of the
lower glucose diffusion rates; further evidence of
it is shown in Fig. 6(b). The proportions of the
mean growth rates over the first 12h (Fig. 6b) with
one, two and three membranes respectively were
2.9:1.7:1.0. Another noteworthy feature of the
lowest glucose feed-rate (curve H) is the apparent
inhibition of growth initially; this indicates that a
marked decrease in the glucose supply rate below
the requirement for maximum growth rate requires
a considerable adaptation in the cells. The culture
with all the glucose present initially (curve E) went
into a decline phase after 12h. In cultures G and H
the growth rates tended to a common minimum
value after 24h; this could be attributed to the fact
that at this stage most of the glucose fed would be
consumed for maintenance energy (Pirt, 1965). At
this point we are confronted with an almost totally
unexplored field of bacterial physiology. The
diffusion-feed technique should greatly facilitate the
investigation of cells supplied with energy source
at or near to the maintenance ration. The decreased
growth yields with slow diffusion (all cultures in the
slow-diffusion experiment had the same amount of
glucose) may in part be due to the greater proportion
of glucose used for maintenance.
Vol. 121
DIFFUSION FEED
The diffusion capsule greatly increases the scope
of the shake-flask technique by providing a simple
means of growing organisms that are not saturated
with substrate and consequently have growth rates
below the maximum. This simulates, in a shakeflask, one of the essential features of continuousflow culture of the chemostat type. Diffusion feed
may be used as a quick method of screening organisms for some of their reactions under the more
elaborate chemostat conditions. Experiments on
diffusion feed of lactose to cultures of E. coli showed
that the fi-galactosidase content of the biomass was
roughly proportional to the lactose feed rate.
The diffusion capsule may be used as a means of
overcoming feed-back inhibition or enzyme repression by restriction of the supply of a nutrient;
this use of the capsule would be analogous to the
more elaborate chemostat method used by Gorini
(1960) to demonstrate the relief of arginine repression of the synthesis of ornithine transcarbamylase.
The stimulation of penicillin production in
cultures of Penicillium by restriction of the glucose
supply, first reported by Soltero & Johnson (1953),
297
is readily achieved in shake-flask cultures by
diffusion feed. Another application for which the
capsule should be suitable is the supply of a toxic
substrate so that its concentration always remains
below the inhibitory value. Also the capsutle should
provide a convenient means for supplying a pulse
of radioactive tracer so that the excess of tracer can
instantly be removed from the culture.
The workshop assistance of Mr K. Reynolds and the
technical assistance of Miss Patricia Adams are gratefully
acknowledged. Patent applications have been made in
the U.K. and other countries and the rights have been
assigned to the National Research Development Corporation.
REFERENCES
Gorini, L. (1960). Proc. natn. Acad. Sci. U.S.A. 46, 682.
Pirt, S. J. (1965). Proc. R. Soc. B, 163,224.
Schultz, J. S. & Gerhardt, P. (1969). Bact. Rev. 33, 1.
Soltero, F. V. & Johnson, M. J. (1953). Appl. Microbiol.
1,52.
Yemm, E. W. & Cocking, E. C. (1955). Analyst, Lond.,
80, 209.