ABSTRACT
Biochemistry is usually carried out under in vitro conditions that do not closely
approximate those found in vivo. Molecular crowding, resulting from high intracellular
protein concentrations, has been shown to affect macromolecular function (Garner & Burg,
1994). To study this effect, the thermal stabilities of mitochondrial and cytoplasmic
isoforms of malate dehydrogenase (MDH), an enzyme involved in the Krebs cycle, were
studied under simulated in vivo conditions of high protein concentration. The
mitochondrion contains much higher protein concentrations than those found in the
cytoplasm; protein crystals have been used to model the mitochondrial matrix (Srere,
1981). The mitochondrial isoform of MDH (mMDH) showed substantial stability
enhancement in media crowded with bovine serum albumin, a model protein. The
cytoplasmic isoform (cMDH), which is more thermally stable than the mitochondrial
isoform, exhibited no si gnificant enhancement of stability in crowded media. This
difference can be seen as an adaptation to local protein concentration. mMDH must be
more flexible to function well in a more crowded medium, rendering it less thermally stable
and more responsive to the stabilizing effects of a crowding agent. In order to study
enzyme function, which is closely related to stability, apparent Michaelis-Menten constants
(Km's) and Vmax's were determined in crowded and viscous conditions for the LDH of
three species. The LDH of Parachaenichthyscharcoti , an Antarctic fish, appears less
susceptible to the effects of agents that slow enzyme function than two homologous
enzymes from organisms that maintain a warmer body temperature. LDH from colder
temperature organisms appears to be more flexible than LDH from organisms adapted to
warm temperatures (Fields and Somero, in prep). This flexibility may explain the results
obtained.
INTRODUCTION
The cell is a very complex environment, but in vitro biochemistry is usually carried out in
dilute buffer solutions with the components of interest highly purified (Garner & Burg,
1994). While this type of simplification is often necessary, it can yield an inaccurate view
of cellular processes. A very dilute solution is thermodynamically ideal, but the cytoplasm
of the cell deviates greatly from ideality (Garner & Burg, 1994). Proteins and small
molecules in solution affect one another’s activities, and while this situation can be modeled
by multicomponent thermodynamics, it is difficult to predict how the activity of a given
macromolecule will be affected (Garner & Burg, 1994).
When a significant part of a solution is taken up by macromolecules, the spaces between
them are no longer large compared to the size of the macromolecules. In such a case,
creating a space for another large molecule is entropically unfavorable because of the higher
order of the resulting solution (Garner and Burg, 1994). Because proteins are flexible and
can sample many possible configurations, this crowding effect may favor a more compact
protein structure. When proteins are exposed to heat, they denature. Because denatured
proteins occupy a greater volume than native proteins, it is plausible that macromolecular
crowding enhances protein stability.
The protein concentration of the cytoplasm is very high, about 20% (Fulton, 1982). The
mitochondrial protein concentration is even higher, about 56% (Srere, 1981). This is
similar to the concentration found in protein crystals, which have been used to model the
mitochondrial environment (Srere, 1981). In addition to having a higher protein
concentration than the cytoplasm, the mitochondrion contains its own genetic code. Some
enzymes have separate mitochondrial and cytoplasmic isoforms. Malate dehydrogenase
(MDH), which mediates the conversion of oxaloacetic acid and NADH to malate and
NAD+ in the Krebs cycle, is one such enzyme. The mitochondrial isoform, mMDH, has
been determined to be much less thermally stable than the cytoplasmic isoform, cMDH,
under in vitro conditions (Somero, pers. comm.). In one phase of this study, it was
determined whether this difference persists under more realistic protein concentrations,
using bovine serum albumin (BSA) as a model protein. Since solutions containing high
concentrations of BSA are more crowded than dilute solutions, it would be reasonable to
expect some degree of stabilization as a result of molecular crowding. Enzymes undergo
configurational changes during catalysis, so it is also reasonable to expect that kinetic
parameters such as Vmax and Km are affected by molecular crowding.
Km and Vmax are theoretical quantities describing the catalytic activity of an enzyme. The
Michaelis-Menten model, the basis for the calculation of these quantities, assumes that an
intermediate in enzymatic action is the formation of an enzyme-substrate complex, which
can either dissociate back into enzyme and substrate or go on to form enzyme and product.
At high concentrations of substrate, the substrate will saturate the enzyme, so the limiting
steps in the reaction will be the ability of the enzyme to undergo a conformational change
and the speed of conversion of substrate to product. Since a further increase in substrate
concentration will not have an effect on these steps, the velocity of the reaction will reach a
maximum, Vmax, as substrate concentration is increased. Km is defined to be the
substrate concentration at which the velocity of the reaction is half of Vmax. Km often is a
measure of the binding affinity for the substrate, the tendency to form the enzyme-substrate
complex. If half-velocity is reached at a low substrate concentration, it means the enzyme
has a high binding affinity for the substrate, i.e., a low Km. If the enzyme requires a lot of
substrate to reach this level, the enzyme has a low affinity for substrate, i.e., a high Km.
Solutions of dextran, a high molecular weight polymer, have been shown to decrease the
apparent Km of lactate dehydrogenase (LDH), an enzyme involved in anaerobic
metabolism (Laurent, 1971). This can be explained by exclusionary effects. With part of
the volume of the medium excluded by the polymer, the activity coefficients of the other
components increase (Laurent, 1971). In a crowded solution, the concentration of
pyruvate, an LDH substrate, is higher in the solution accessible to it than it would be in
pure water. LDH has more substrate available to it, and hence has a lower Km in the
presence of dextran (Laurent, 1971). The present study used polyethylene glycol,
molecular weight 8000 (PEG-8000), to study the effect of a large molecule on KmPYT and
Vmax. An additional area of study involved the effect of glycerol on LDH kinetics. Since
glycerol is a small molecule like water, it does not produce exclusion effects, so its
influence on enzymatic function is not due to molecular crowding. Glycerol is miscible
with water and, like PEG-8000, its solutions are more viscous than water. Although
glycerol solutions are less viscous than PEG-8000 solutions with the same concentration,
glycerol could provide a way to study viscosity effects on enzyme function which are
independent of molecular crowding.
Evidence has accumulated that water itself is an important factor influencing conformational
changes in molecules. It has been shown that 60 water molecules bind to hemoglobin as it
goes from the deoxygenated to the oxygenated state (Colombo et al., 1992). Rand et al.
(1992) urge one "not to ignore water as a ligand in allosterism." The activity of water
varies depending on the constituents of a solution, and effects on ligand binding can be due
solely to water. Rand et al. (1992) found that Km of glucose for hexokinase depended
only on the activity of water, which was "adjusted with high molecular weight osmolytes.
It was also found that glucose binding entails the release of 65+10 water molecules (Rand
et al., 1992). When macromolecules are present, water tends to arrange itself around them
in layers. Much of water in the cell is likely to be found in this bound state, although the
exact extent is a matter of some controversy. The difference in behavior between this water
and free (unbound) water, is responsible for changes in the activity of water in the presence
of macromolecules.
Viscosity, which increases significantly with decreases in temperature, has been shown to
be an important variable in physiology. 40% of the reduction in swimming speed and 55%
of the reduction in water movement of the larvae of the sand dollar Dendraster excentricus
with a decrease in temperature were found to be due to the increased viscosity of the cooler
medium, rather than temperature per se (Podolsky and Emlet, 1993). At the biochemical
level, Beece et al. (1980) state that "viscosity may play an essential role in determining the
state of the protein through structural fluctuations." High viscosity inhibits the ability of
proteins to move through a variety of substates (Beece et al., 1980). Surprisingly, the free
water of the cell possesses a viscosity only 1.2-1.4 times that of pure water at the same
temperature (Fushimi and Verkman, 1991). Much of the Q10 effect of temperature on
enzymatic activity may be explainable by viscosity, just as it explained some of the effect
on Dendraster excentricus. This question was investigated in this study. In addition,
crowding and viscosity effects on the LDH of different species were compared.
LDH from Gillichthys mirabilis(the long-jawed mudsucker), Parachaenichthyscharcoti (an
Antarctic fish), and rabbit, was used in this experiment. The AA isoform, which is found
in muscle rather than heart, and favors the conversion of pyruvate and NADH to lactate and
NAD“, was used. G. mirabilis is a eurythermal fish and lives in temperate waters. P.
charcoti lives near the freezing point in Antarctic waters, and rabbits maintain body
temperature at around 40°C. Since the viscosity of water varies from 1.8 cp. at 0°C to 0.65
cp. at 40°C, it would not be surprising to see some adaptation to viscosity found in the
LDH of these species. This was examined in this experiment by gathering A4-LDH kinetic
data in viscous conditions.
MATERIALS & METHODS
Thermal Denaturation of MDH
MDH samples were prepared from cytoplasmic and mitochondrial porcine heart MDH
available commercially from Sigma Chemical Company. To remove the suspending agent,
ammonium sulfate, the MDH was concentrated in a centricon at 700 g using 20 mM
’C. Samples to be denatured contained 70-75 mM
imidazole chloride buffer (pH 7.0 at 2
imidazole chloride (pH 7.0 at 22°C), varying concentrations of BSA, and MDH diluted
approximately 1000 fold from the original commercial sample. Samples were heated in a
thermal cycler for the appropriate time interval. They were kept on ice for at least five
minutes, and then centrifuged at 15,800 g for 5 minutes to remove aggregated protein.
Triplicate enzyme activity assays were performed on the supernatant.
Assay of Enzyme Activity
MDH or LDH activity was determined by measuring the decrease in absorbance at 340 nm,
the absorbance maximum of NADH, per unit time, using a Shimadzu Bio-Spec 1601 UV¬
Vis Spectrophotometer. The MDH assay medium contained 80 mM imidazole chloride (pH
7.0 at 22°C), 120 uM NADH, and 200 pM oxaloacetic acid. The LDH assay medium
contained 80 mM imidazole chloride (pH 7.0 at 22°C), 150 M NADH, and 2 mM
pyruvate. The reaction temperature was maintained at 20°C.
Native Polyacrylamide Gel Electrophoresis (PAGE)
To determine if there was more than one source of MDH activity in the cMDH solutions, a
native PAGE assay was performed. This assay separates proteins by charge and size, and
uses a stain to detect MDH activity. The procedure from Brewer (1970) was followed.
Determination of MDH Half-lives
The log of MDH residual activity was plotted against time, and the best fit line was
obtained using Kaleidagraph software. Because cMDH did not closely follow an
exponential decay curve, the initial activity measurements (activity before denaturation)
were not used in the calculation of cMDH half-lives. These points deviated most strongly
from the exponential decay curve established by the other points.
Km and Vmax calculation
A4-LDH KmPYf and Vmax values were determined using an assay medium containing 80
mM imidazole chloride (pH 7.0 at 22°C), 150 uM NADH, and different concentrations of
pyruvate (0.05, 0.075, 0.1, 0.15, 0.2, 0.3, 0.4, 0.5, and sometimes 0.8 mM). The
medium also contained the appropriate concentration of glycerol or PEG-8000 if these were
necessary for the assay. Triplicate assays were performed for each pyruvate concentration.
KmPVT and Vmax were calculated using the method of Wilkinson (1961) and Wilman
software (Brooks and Suelter, 1986). Those points that deviated from the calculated
regression line by more than two standard deviations were eliminated. Each calculated
Vmax was correlated to others determined on different days by measuring LDH activity
under saturation conditions: a medium containing 150 uM NADH, 2 mM pyruvate, and 80
mM imidazole chloride (pH 7.0 at 22°C).
Viscosity Measurements
Viscosities of PEG-8000 and glycerol solutions were determined by using a Gilmont
Instruments Falling-Ball Viscometer.
RESULTS
Thermostability measurements indicated that porcine mMDH stability was enhanced in the
presence of BSA, with successive increases in the half-life of mMDH at 42°C with each
increase in BSA concentration (Figs. 1, 2,3). The decay curve of mMDH was exponential
(Fig. 2). CMDH at 42° was more stable than mMDH at 42° (Fig. 3). CMDH did not follow
an exponential decay curve as well as mMDH (Figs. 4, 5), and its stability was not
significantly enhanced by high BSA concentrations (Figs. 4, 5). When a native PAGE
assay was performed, a small secondary band was seen in the cMDH sample. This band
disappeared upon heat denaturation of the cMDH sample.
High BSA concentrations produced reductions in the catalytic activity of porcine mMDH
(Fig. 6). Glycerol and PEG-8000 solutions of increasing concentration produced
progressively greater reductions in the catalytic activity of AA-LDH (Figs. 7, 8). P.
charcoti AA-LDH was affected less by high glycerol concentrations than rabbit or G.
mirabilis AA-LDH (Fig. 7). PEG-8000 solutions had higher viscosities than glycerol
solutions at the same concentration (Fig. 9). 20% glycerol solutions produced reductions
in both Vmax and KmPVT of A4-LDH of all three species (Tables 1,2). 3.5% PEG-8000
solutions, which have about the same viscosity as 20% glycerol solutions, produced
smaller reductions in Vmax and KmPYf (Tables 1,2) Temperature reduction from 20°C to
10°C produced reductions in both Vmax and KmPYT (Tables 1, 2).
DISCUSSION
Enhancement of mMDH stability in progressively higher protein concentrations is
consistent with developed models of molecular crowding (Garner & Burg, 1994). High
concentrations of BSA favor more compact protein structures, which may be less prone to
denaturation. This study suggests that these effects can be very large, judging by the
enhancement of mMDH stability in the presence of high BSA concentrations. mMDH
probably shows much more stability enhancement than cMDH because it adopts a looser
and more open conformation in dilute, in vitro, conditions. BSA forces this loose
conformation into a more compact state, decreasing the likelihood of denaturation. CMDH
is probably already so compact and stable that addition of a crowding agent has little effect
on its conformation or its thermal stability. The lower stability of mMDH is probably an
adaptation to the high protein concentration of the mitochondrion. It has been postulated
that there is a tradeoff between stability and function in enzymes (Somero, 1995). With a
loose, open conformation, mMDH is more likely than cMDH to be flexible enough to
catalyze effectively at high protein concentrations. This study showed that much of the gap
in apparent thermal stability between cytosolic and mitochondrial isoforms is due to
differences in the protein concentrations of the cytosol and mitochondria. Extrapolating the
mMDH curve of Fig. 3 out to 500 mg/ml, the approximate protein concentration of the
mitochondrion (Srere, 1981), and the cMDH curve out to 200 mg/ml, the approximate
cytosolic protein concentration (Fulton, 1982), there is still a difference in thermal stability
between the isoforms, but it is much smaller than the difference under dilute conditions.
The flattening of the mMDH curve near 100 mg/ml (Fig. 3) is probably due to the adoption
of a substantially compacted state, which is more or less immune to additional crowding
effects. This is likely the state that cMDH already exists in at 42°, even at low BSA
concentrations. Fig. 4 shows the similarity between the denaturation curves of cMDH in
Img/mi BSA and 75 mg/ml BSA at 42°C.
This present study also examined molecular crowding effects on enzyme kinetics, the other
half of the stability versus function equation. Consistent with a model of enzyme activity in
which a more compact, less flexible enzyme configuration results in a lower rate of
catalysis, 3.5% PEG-8000 solutions produced small reductions in catalytic activity in the
three species studied (Table 1). 20% PEG-8000 produced a large drop in the catalytic
10
activity of G. mirabilis AA-LDH at 20° (Table 1), but this medium was not used for other
species or temperatures because its high viscosity (about 19 times that of water) made it
difficult to work with. By crowding A4-LDH, PEG-8000 solutions likely make it difficult
for AA-LDH to undergo the full range of motion necessary for function, thereby slowing
catalysis and lowering Vmax values. This also explains BSA’s effect on mMDH activity
(Fig. 6). PEG-8000 molecules are too large to fit into the binding cleft of A4-LDH, so
PEG-8000's action is probably not due instead to direct interference. 20% glycerol
solutions, which have about the same viscosity as 3.5% PEG-8000 solutions (2.2 cp. at
), produced much larger drops in catalysis rates than 3.5% PEG-8000 did (Table 1).
22
Glycerol may act by getting into the binding cleft and slowing catalysis, as it is a small
molecule. Dunn et al. (1991) postulated that solvent friction slows enzymes. When LDH
closes upon the substrates during binding, the water molecules in the vacuole become
highly ordered (Dunn et al., 1991). With glycerol present, this process will be affected, so
glycerol could act in this fashion. P. charcoti AA-LDH retained more activity than rabbit or
G. mirabilis A4-LDH in the presence of glycerol or PEG-8000 (Table 1). Since P.charcoti
resides in Antarctic waters, which as a result of their temperature have a viscosity 1.8 times
that of water at 20°C, this result may be due to adaptation to a high viscosity medium. It
could be due to the presence of a population of P.charcoti LDH molecules whose
conformations are too "loose" to be functional at high temperatures or low viscosities, but
which become functional upon a drop in temperature or introduction into a more viscous
medium (P. Fields, pers. comm.). Enzymes from cold-temperature organisms are thought
to be more flexible (Fields and Somero, in prep), and since enzymes continually sample
small configurational changes, there will be a certain population of molecules which at a
given moment are "looser" than others. At 20° and no solute present, the reference point
for the Vmax calculations, some of the P.charcoti molecules may be nonfunctional, but all
the G. mirabilis and rabbit AA-LDH molecules are likely functional, as they normally exist
at higher temperatures.
11
If one assumes that glycerol’s effect on AA-LDH kinetics is primarily due to viscosity, it
follows that some of the Q1O effect is due to viscosity per se. 20% glycerol solutions
lowered the Vmax of A4-LDH significantly (Table 1). Viscosity effects on enzymes
deserve much more investigation.
Conditions that resulted in lower Vmax's produced a concomitant decrease in apparent Km
(Tables 1,2). This may be due to stabilization of the binding-competent state of the
enzyme (Somero, pers. comm.). When an enzyme is slowed down, it samples fewer
configurations, and may be more likely to exist in a state in which substrate is bound. This
results in higher affinity for substrate, i.e., a lower Km. This is speculative, however.
Laurent (1971) has already shown that polymers reduce apparent Km's by reducing water
activity, thus explaining the Km decreases seen with PEG-8000.
Thermodynamics can often make good predictions of how enzymes will be affected by
different media (Garner & Burg, 1994). However, it cannot readily predict differences
between homologues of an enzyme. In addition, the Michaelis-Menten model may be an
inadequate model of how enzymes actually work (Somero, pers. comm.). More work is
needed in this area
Simplified models tend to belie the complexity of what is actually happening in vivo.
Studying biochemical systems isolated from their surrounding is like studying an organism
taken from its natural habitat. Although this may be helpful in understanding many aspects
of the organism, this type of work cannot adequately describe how the organism will
behave in its natural environment. An ecological approach to biochemistry, in which
cellular components are studied in conditions which more closely represent their natural
environment, should provide many more insights.
ACKNOWLEDGEMENTS
I wish to thank George Somero and Peter Fields for their invaluable assistance in
completing this study. I also with to thank Lars Tomanek, Jonathon Stillman, and Andy
Gracey for their help.
LITERATURE CITED
Beece et al. 1980. Solvent viscosity and protein dynamics. Biochemistry. 19:5147-
5157.
Brewer, G.J. 1970. pp. 115-116 in Introduction to isozyme techniques. Academic Press,
New York.
Brooks, S.P.J. and C.H. Suelter. 1986. Estimating enzyme kinetic parameter
computer program for linear regression and nonparametric analysis. Int. J. bio-med.
Computing. 19:89-99.
Colombo, M.F., D.C Rau and V.A. Parsegian. 1992. Protein solvation in allosteric
regulation: A water effect on hemoglobin. Science. 256:655-659.
Dunn et al. 1991. Design and synthesis of new enzymes based on the lactate
dehydrogenase framework. Phil. Trans. R. Soc. Lond. 332:177-184.
Fulton, A.B. 1982. How Crowded is the Cytoplasm? Cell. 30:345-347.
Fushimi, K and A.S. Verkman. 1991. Low viscosity in the aqueous domain of cell
cytoplasm measured by picosecond polarization microfluorimetry. J. Cell Bio.
112:719-725.
Garner, M.M. and M.B. Burg. 1994. Macromolecular crowding and confinement in cells
exposed to hypertonicity. Am. J. Physiol. 266:C877-C892.
Laurent, T. 1971. Enzyme reactions in polymer media. Eur. J. Biochem. 21:498-506.
Podolsky, R.D. and R.B. Emlet. 1993. Separating the effects of temperature and
viscosity on swimming and water movement by sand dollar larvae (Dendraster
excentricus). J. exp. Biol. 176:207-221.
Rand, R.P. 1992. Raising water to new heights. Science. 256:618.
Somero, G.N. 1995. Proteins and Temperature. Annu. Rev. Physiol. 57:43-68.
Srere, P.A. 1981. Protein Crystals as a Model for Mitochondrial Matrix Proteins.
Wilkinson, G.N. 1961. Statistical estimations in enzyme kinetics. Biochem. J. 80:323-
332.
13
20
Table 1: Vma
Values. Numbers Represent
Percentage of the Vpax at 20°C with no Solute
Present for That Species. Errors are Standard
Deviations.
No solute
3.5% PEG-8000 20% PEG-8000
20% glycerol
49(41)
47(42)
G. mirabilis
24.9(40.4)
charcoti
54(41)
53(41)
39.6(40.5)
54(41)
48(41)
rabbit
31(41)
48(41)
G. mirabilis
100(43)
88(42)
59(43)
charcoti
100(42)
73(42)
99(41)
rabbit
100(41)
66(41)
94(42)
20
pyr in mM. Errors are
Table 2: Apparent A,-LDH K
Standard Deviations.
3.5% PEG-8000
No solute
20% glycerol
20% PEG-8000
O.139(+0.006)
G. mirabilis
O.050(40.003
O.110(+0.009)
P. charcoti
0.28(40.01)
0.153(40.004
0.24(40.01)
rabbit
0.09(40.01)
0.140(40.008
0.062 (40.003)
G. mirabilis
0.20(40.01)
O.154(40.007)
0.06(40.01)
0.062 (40.003)
0.41(40.01)
P. charcoti
0.51(40.02)
0.24(40.01)
0.143(40.006)
rabbit
0.158(40.005)
0.093(40.003)
FIGURE LEGENDS
Fig. 1. Activity of porcine mMDH after denaturation at 42°C in 6 different concentrations
of BSA. Error bars are standard deviations from the mean, n =3 for each point.
Fig. 2. Activity of porcine mMDH after denaturation at 42°C in 6 different concentrations
of BSA. y-axis is logarithmic to show exponential decay pattern. Error bars are standard
deviations from the mean, n=3 for each point.
Fig. 3. Half-lives of MDH in media containing different concentrations of BSA. Error
bars are standard errors.
Fig. 4. Activity of porcine cMDH after denaturation at 42°C in 2 different concentrations of
BSA. Error bars are standard deviations from the mean, n-3 for each point.
ig. 5. Activity of porcine cMDH after denaturation at 48°C in 3 different concentrations o
BSA. Error bars are standard deviations from the mean, n =3 for each point.
Fig. 6. Activity of porcine mMDH at 20° in media containing different concentrations of
BSA. Error bars are standard deviations from the mean, n-3 for each point.
Fig. 7. A4-LDH activity at 20° as a function of glycerol concentration for three different
species. Error bars are standard deviations from the mean, n =3 for each point.
Fig. 8. A4-LDH activity at 20° as a function of PEG-8000 concentration for two species.
Error bars are standard deviations from the mean, n- 3 for each point.
Fig. 9. Viscosity at 22°C as a function of glycerol and PEG-8000 concentration.
Fig. 1
—

100

1 mg/mi BSA
10 mg/mi BSA
38 mg/mi BSA
80
75 mg/mI BSA
93 mg/mi BSA
150 mg/mi BSA
60
40
20

40 80 120 160 200 240 280
Time (min.)
E
Fig. 2
tk kkk-
k kka kk vak-
——
100
90
1 mg/mi BSA
80
10 mg/mi BSA
38 mg/mi BSA
70
75 mg/mi BSA
93 mg/mi BSA
60
150 mg/mI BSA
I
30

40 80 120 160 200 240 280
Time (min.)
E
10
1000
100
Fig. 3
mMDH Q 42°
CMDH Q 42°0
A CMDH Q 48°0



20 40 60 80 100 120 140 160
BSA conc. (mg/ml)
E
Fig. 4
100
90
1 mg/mi BSA
75 mg/mI BSA

80
70
60
50
40
—
200 400 600 800 1000 1200 1400
Time (min.)
100
50
Fig. 5
—
100
150
Time (min.)
1 mg/mi BSA
10 mg/mi BSA
38 mg/mi BSA
200
250
24
S0

5
100
80
60
20
ig. 6
4
L
20
40
60
80
100
BSA Concentration (mg/ml)
Fig. 7

0
A G. mirabilis
P. charcoti
rabbit
—80

5-
8
2
2 40
2

20
E
40
10
20
30
50
Glycerol Concentration (% V/V)
Fig. 8
kk k t-

1004
A
A
80
A P. charcoti
60
58
G. mirabilis
—
u0
OL
40
20
E
10
15
20
PEG-8000 Concentration (% V/V)
3
Fig. 9
1aaaaa-
ka kkak vakakava-
glycerol


- PEG-8000
L

10 20
40 50
60
Solute conc. (% V/V)