Effect of chemical chaperones in improving the solubility of recombinant proteins in Escherichia coli.
ABSTRACT The recovery of active proteins from inclusion bodies usually involves chaotrope-induced denaturation, followed by refolding of the unfolded protein. The efficiency of renaturation is low, leading to reduced yield of the final product. In this work, we report that recombinant proteins can be overexpressed in the soluble form in the host expression system by incorporating compatible solutes during protein expression. Green fluorescent protein (GFP), which was otherwise expressed as inclusion bodies, could be made to partition off into the soluble fraction when sorbitol and arginine, but not ethylene glycol, were present in the growth medium. Arginine and sorbitol increased the production of soluble protein, while ethylene glycol did not. Production of ATP increased in the presence of sorbitol and arginine, but not ethylene glycol. A control experiment with fructose addition indicated that protein solubilization was not due to a simple ATP increase. We have successfully reproduced these results with the N-terminal domain of HypF (HypF-N), a bacterial protein which forms inclusion bodies in Escherichia coli. Instead of forming inclusion bodies, HypF-N could be expressed as a soluble protein in the presence of sorbitol, arginine, and trehalose in the expression medium.
- [show abstract] [hide abstract]
ABSTRACT: Green fluorescent protein (GFP) is an unusually stable autofluorescent protein that is increasingly being exploited for many applications. In this report, we have used fluorescence spectroscopy to study the effect of pH on the denaturation of GFP with sodium dodecyl sulfate (SDS), urea, and heat. Surprisingly, SDS (up to 0.5%) did not have any significant effect on the fluorescence of GFP at pH 7.5 or 8.5 buffers; however, at pH 6.5, the protein lost all fluorescence within 1 min of incubation. Similarly, incubation of GFP with 8 M urea at 50 degrees C resulted in time dependent denaturation of GFP, but only in pH 6.5 buffer. At higher pH values (pH 7.5 and pH 8.5), the GFP was quite stable in 8 M urea at 50 degrees C, showing only a slight decrease in fluorescence. Heat denaturation of GFP was found to be pH dependent as well, with the denaturation being fastest at pH 6.5 as compared to pH 7.5 or pH 8.5. Like the denaturation studies, renaturation of heat-denatured GFP was most efficient at pH 8.5, followed by pH 7.5, and then pH 6.5. These results suggests that GFP undergoes a structural/stability shift between pH 6.5 and pH 7.5, with the GFP structure at pH 6.5 being very sensitive to denaturation by SDS, urea, and heat.Applied Biochemistry and Biotechnology 09/2005; 126(2):149-56. · 1.89 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: Spontaneous mutants of Escherichia coli able to grow on ethylene glycol as a sole source of carbon and energy were obtained from mutants that could grow on propylene glycol. Attempts to obtain ethylene glycol-utilizing mutants from wild-type E. coli were unsuccessful. The two major characteristics of the ethylene glycol-utilizing mutants were (i) increased activities of propanediol oxidoreductase, an enzyme present in the parental strain (a propylene glycol-positive strain), which also converted ethylene glycol into glycolaldehyde; and (ii) constitutive synthesis of high activities of glycolaldehyde dehydrogenase, which converted glycolaldehyde to glycolate. Glycolate was metabolized via the glycolate pathway, which was present in the wild-type cells; this was indicated by the induction in ethylene glycol-grown cells of glycolate oxidase, the first enzyme in the pathway. Glycolaldehyde dehydrogenase was partially characterized as an enzyme of this new metabolic pathway in E. coli, and glycolate was identified as the product of the reaction. This enzyme used NAD and NADP as coenzymes, although the NADP-dependent activity was about 10 times lower than the NAD-dependent activity. Uptake of [14C]ethylene glycol was dependent on the presence of the enzymes capable of metabolism of ethylene glycol. Glycolaldehyde and glycolate were identified as intermediate metabolites in the pathway.Journal of Bacteriology 02/1983; 153(1):134-9. · 3.19 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: A protein determination method which involves the binding of Coomassie Brilliant Blue G-250 to protein is described. The binding of the dye to protein causes a shift in the absorption maximum of the dye from 465 to 595 nm, and it is the increase in absorption at 595 nm which is monitored. This assay is very reproducible and rapid with the dye binding process virtually complete in approximately 2 min with good color stability for 1 hr. There is little or no interference from cations such as sodium or potassium nor from carbohydrates such as sucrose. A small amount of color is developed in the presence of strongly alkaline buffering agents, but the assay may be run accurately by the use of proper buffer controls. The only components found to give excessive interfering color in the assay are relatively large amounts of detergents such as sodium dodecyl sulfate, Triton X-100, and commercial glassware detergents. Interference by small amounts of detergent may be eliminated by the use of proper controls.Analytical Biochemistry 06/1976; 72:248-54. · 2.58 Impact Factor
APPLIED AND ENVIRONMENTAL MICROBIOLOGY, July 2011, p. 4603–4609
Copyright © 2011, American Society for Microbiology. All Rights Reserved.
Vol. 77, No. 13
Effect of Chemical Chaperones in Improving the Solubility of
Recombinant Proteins in Escherichia coli?†
Shivcharan Prasad, Prashant B. Khadatare, and Ipsita Roy*
Department of Biotechnology, National Institute of Pharmaceutical Education and
Research (NIPER), Sector 67, S.A.S. Nagar, Punjab 160 062, India
Received 23 April 2011/Accepted 26 April 2011
The recovery of active proteins from inclusion bodies usually involves chaotrope-induced denaturation,
followed by refolding of the unfolded protein. The efficiency of renaturation is low, leading to reduced yield of
the final product. In this work, we report that recombinant proteins can be overexpressed in the soluble form
in the host expression system by incorporating compatible solutes during protein expression. Green fluorescent
protein (GFP), which was otherwise expressed as inclusion bodies, could be made to partition off into the
soluble fraction when sorbitol and arginine, but not ethylene glycol, were present in the growth medium.
Arginine and sorbitol increased the production of soluble protein, while ethylene glycol did not. Production of
ATP increased in the presence of sorbitol and arginine, but not ethylene glycol. A control experiment with
fructose addition indicated that protein solubilization was not due to a simple ATP increase. We have
successfully reproduced these results with the N-terminal domain of HypF (HypF-N), a bacterial protein which
forms inclusion bodies in Escherichia coli. Instead of forming inclusion bodies, HypF-N could be expressed as
a soluble protein in the presence of sorbitol, arginine, and trehalose in the expression medium.
Overexpression of proteins in heterologous systems often
results in the formation of inclusion bodies. These electron-
refractile particles are formed because of the failure of pro-
karyotic folding machinery to correctly fold the nascent poly-
peptide chain, increasing its local concentration in the cytosol.
The most common strategy to recover active proteins from
these particles is by denaturing them in the presence of chao-
tropes and refolding them to the native conformation by the
gradual removal of the denaturant. This process results in high
loss in protein yield since the renaturation efficiency of most
proteins is not very high. During this process, the protein may
start to reaggregate once an intermediate concentration of the
denaturant is reached. Various strategies have been devised to
overcome this drawback. Refolding of urea-denatured proteins
has been carried out using size exclusion chromatography (24,
36). Attempts have also been made to increase the fraction of
soluble protein expressed in Escherichia coli by decreasing the
growth temperature to between 28 and 30°C (32). Changing
other growth conditions, such as components of the culture
medium, concentration of inducer, and induction time, can
help in increasing the yield of soluble overexpressed protein.
Use of solubilizing fusion tags, coexpression of chaperones,
site-directed mutagenesis, etc., has also been proposed. It has
been suggested that though aggregated, the protein in the
inclusion bodies may be in its native conformation and func-
tionally active (10, 14, 17). In these cases, the protein is easier
to recover since the overall tertiary fold of the protein is not
lost during aggregation. However, the final yield of the active
protein in these cases is variable.
It had earlier been assumed that bacterial inclusion bodies
are amorphous structures with no particular morphology; the
overexpressed protein was thought to form inclusion bodies
because its three-dimensional fold was destabilized. It is, how-
ever, now increasingly seen that protein aggregates in inclusion
bodies have defined structures. In many cases, these have been
shown to form extended ?-amyloid-like sheets (6, 37). Purified
inclusion bodies have been shown to enhance the fluorescence
intensity of thioflavin T and produce apple green birefringence
when stained with Congo red (37) which are characteristic
properties of crossed ?-sheet structures. H/D exchange in1H-
showed discrete regions of strongly ordered structures within
the inclusion bodies, where the exchange rate was very low
(10?3to 10?4h?1) (37).
Osmolytes and pseudochaperones have proven to be useful
in inhibiting protein aggregation during refolding of unfolded/
misfolded proteins in vitro. Incorporation of osmolytes such as
amino acids, sugars, and low concentrations of chaotropes in
the renaturation buffer results in high refolding yield of recom-
binant proteins. Arginine has been one of the most commonly
used folding aids for recovery of soluble proteins from inclu-
sion bodies (35, 40). In all these cases, overexpression of pro-
tein involves formation of inclusion bodies from which the
soluble protein can be recovered using multiple steps. The
initial step of denaturation with chaotropes is followed by a
refolding step. It is at this step that the recovery efficiency
decreases drastically. Attempts to solubilize proteins by de-
creasing the growth temperature have resulted in lower protein
yield, which has an undesirable effect on the protein produc-
tion process. In this work, we report that green fluorescent
protein (GFP) and the N-terminal domain of HypF (HypF-N),
which were expressed as inclusion bodies under normal growth
15N-crossed nuclear magnetic resonance spectra clearly
* Corresponding author. Mailing address: Department of Biotech-
nology, National Institute of Pharmaceutical Education and Research
(NIPER), Sector 67, S.A.S. Nagar, Punjab 160 062, India. Phone: 91
172 229 2061. Fax: 91 172 221 4692. E-mail: firstname.lastname@example.org.
† Supplemental material for this article may be found at http://aem
?Published ahead of print on 6 May 2011.
conditions, partitioned off into the soluble fraction when grown
in the presence of osmolytes at 37°C, without any decrease in
the final protein yield. We have attempted to understand the
mechanism by which the transition from insoluble to soluble
phase occurred in the presence of osmolytes. The mechanism
appears to be complex. The extent of solubilization seems to
correlate well with the ATP content of the cells, which in-
creased in the presence of osmolytes, although other factors
are also in operation.
MATERIALS AND METHODS
Expression of GFP. Escherichia coli BL21(DE3) cells were made competent by
the chemical method using calcium chloride (31) and were transformed with the
plasmid pBAD.GFPwt599. The details of the plasmid construct have been de-
scribed earlier (9). For expression of GFP, transformed cells were plated on
LB-ampicillin (LBamp) plates in the absence and presence of 0.2% (wt vol?1)
arabinose and 0.2% (wt vol?1) and 0.4% (wt vol?1) glucose and incubated at
37°C. The fluorescent colonies were visualized under a handheld UV lamp set at
The transformed cells were also grown in liquid culture medium. For this,
single ampicillin-resistant colonies were picked from the plates and grown in
LBampstarter medium (2.5% [wt vol?1] 0.006% ampicillin, in distilled water or
0.1 M phosphate buffer, pH 7.8, 200 rpm, 37°C) overnight and then subcultured
in the same medium until the A600reached 0.2. In each case, 100 ml culture
medium was added to 500-ml nonbaffled flasks. Expression of GFP was induced
by the addition of 0.2% (wt vol?1) arabinose. Cells were grown for 8 h and
harvested by centrifugation at 6,500 ? g for 10 min at 4°C.
Isolation of inclusion bodies. The cell pellet was lysed using lysis buffer (0.05
M Tris-HCl, pH 8.0, 0.001 M EDTA, 0.1 M NaCl) containing lysozyme (5 mg
ml?1). The cell lysate was clarified by centrifugation at 18,000 ? g for 25 min.
The pellet was resuspended in inclusion body solubilization buffer (lysis buffer
containing 8 M urea) for 1 h. This suspension was diluted 10 times with lysis
buffer. The fluorescence intensities of the cell lysate and urea-solubilized fraction
were measured at 509 nm using an excitation wavelength of 395 nm (RF5301;
Shimadzu). The excitation and emission slit widths used were 5 nm each.
Solubilization of GFP. Expression of GFP was induced in culture flasks by the
addition of arabinose (to a final concentration of 0.2% [wt vol?1]) in the pres-
ence of different concentrations of osmolytes. Cells were grown for 8 h, harvested
by centrifugation at 6,500 ? g for 10 min at 4°C, and lysed as described above.
The final volume of lysis buffer was equal to the volume of refolded urea-
solubilized fraction obtained after induction with 0.2% (wt vol?1) arabinose
Estimation of protein content. The amount of soluble protein in the cleared
lysate was determined by the dye binding method (5), using bovine serum
albumin as a standard protein.
Detection of sorbitol, trehalose, and glycerol. Quantitative determination of
polyols in different cell lysates was carried out by high-performance liquid chro-
matography (HPLC; SCL-10A VP; Shimadzu) (12, 28). For estimation of intra-
cellular polyols, cell pellets were suspended in 3 volumes of 0.5 M trichloroacetic
acid and extracted for 40 min at room temperature (25). The extraction was
repeated once, and the two extracts were pooled and analyzed by HPLC. The
clarified samples were loaded on a carbohydrate column (Zorbax; Agilent Tech-
nologies) and detected using a refractive index detector (RID-10A; Shimadzu).
Acetonitrile-water (70:30, vol vol?1) was used as the mobile phase at a flow rate
of 1 ml min?1. Standard curves were plotted using different known concentra-
tions of sorbitol, trehalose, and glycerol, which were used to estimate the
amounts of polyols present in the samples, depending on differences in retention
Detection of arginine. Determination of the arginine content of the cell lysate
was carried out as already described (29). Briefly, equal volumes of test samples
and 3 N NaOH were mixed, a developing solution containing 5% (wt vol?1)
?-naphthol and 2.5% (vol vol?1) diacetyl in water was added, and the mixture
was incubated for 30 min at room temperature. The absorbance of the solution
was measured at 535 nm.
Determination of ATP content. ATP content of different bacterial cell samples
was estimated using a BacTiter-Glo microbial cell viability assay kit (Promega).
Briefly, transformed cells (containing either the empty vector or the GFP insert)
were grown in the presence of the inducer and different osmolytes. After 8 h, the
cells were harvested. A total of 105cells were taken in a total volume of 100 ?l
each and pipetted into wells of opaque-walled multiwell plates. ATP analysis was
carried out according to the manufacturer’s instructions. A standard curve was
plotted with known amounts of ATP provided by the manufacturer. The lumi-
nescence due to luciferase was measured using a Microbeta Trilux luminescence
counter (Perkin Elmer). In order to confirm the noninterference of inducer or
other medium components in the measurement of luminescence due to ATP, an
empty vector, without the sequence of the GFP insert, was constructed. Excision
of the GFP insert (700 bp) from pBAD.GFPwt599 (5.4 kbp) was carried out
using restriction digestion with EcoRI and NheI, according to the manufacturer’s
protocol (Fermentas). The completion of digestion was confirmed by agarose gel
(1%, wt vol?1) electrophoresis. The digested vector backbone (pBAD without
the GFP insert, 4.7 kbp) was purified by preparative gel electrophoresis, followed
by isolation using a Wizard SV gel and PCR cleanup system (Promega, Madison,
WI). Ligation of the double-digested vector backbone was carried out using two
oligonucleotide sequences, AATTCATATATG (containing the partial EcoRI
recognition site, in italics) and CTAGCATATATG (containing the partial NheI
recognition site, in italics), in the presence of T4 DNA ligase (Fermentas) at 16°C
overnight. This empty vector was transformed into chemically competent E. coli
BL21(DE3) cells, which were plated on an LBampplate and grown at 37°C
overnight. A single ampicillin-resistant colony was taken and grown in liquid
LBampmedium. Inducer (0.2% [wt vol?1] arabinose) was added at an A600of 0.2.
The absence of GFP was confirmed by spectrofluorimetry since no peak was
observed at 509 nm, the emission maximum of the protein.
Expression of HypF-N. Escherichia coli BL21(DE3) cells were made compe-
tent by the chemical method using calcium chloride (31) and transformed with
the plasmid HypF-N-His6. The DNA encoding HypF-N (residues 1 to 91) was
cloned from the E. coli K-12 strain (GenBank accession number AP009048) into
the pET21a(?) vector with a C-terminal His6tag by PCR and confirmed by
sequencing (L. Wang, personal communication). Expression of HypF-N was
induced by the addition of 1 mM isopropyl-?-D-thiogalactopyranoside (IPTG).
Cells were grown for 3 h and harvested by centrifugation at 6,500 ? g for 10 min
at 4°C. Cells were also grown in the presence of different osmolytes and pro-
cessed as described above. The presence of soluble HypF-N was monitored by
Western blotting using anti-His6monoclonal antibody as the primary antibody
and horseradish peroxidase (HRP)-conjugated antimouse polyclonal antibody as
the secondary antibody. Bands were visualized using tetramethylbenzidine-H2O2
as the substrate for HRP.
Competent E. coli BL21(DE3) cells, transformed with
pBAD.GFPwt599, were grown in LBampmedium. Expression
of GFP was induced with 0.2% (wt vol?1) arabinose. The
expression of GFP was confirmed by plating the transformed
cells on selection plates (Fig. 1A). Arabinose was found to
induce the expression of the GFP gene since the latter had
been fused to the araC operon. The presence of glucose in the
medium reduces the concentration of cyclic AMP (cAMP) and
hinders the formation of cAMP-cyclic AMP receptor protein
(CRP) complex, inhibiting the activation of the araC operon
(33). When transformed E. coli cells were plated on a selection
medium containing the inducer and glucose at two different
concentrations (0.2% and 0.4%, wt vol?1), the fluorescence of
the colonies formed on the plate was found to decrease (Fig.
1A), indicating that the expression of GFP was inhibited by the
presence of glucose (Fig. 1A). After cell lysis, the presence of
GFP in the soluble and insoluble fractions was monitored by
measuring the fluorescence intensities of the samples. Figure
1B shows the spectrum obtained when the fluorescence inten-
sity of the cell lysate was monitored. The fluorescence intensity
of GFP in the soluble fraction was quite low. The cell pellet
was solubilized in 8 M urea and then diluted 10-fold with lysis
buffer so that the final concentration of urea was 0.8 M. Under
these conditions, most of the fluorescence intensity could be
detected in the cell pellet (Fig. 1B). Thus, GFP is expressed as
inclusion bodies in E. coli BL21(DE3) cells. In the initial ex-
periments, the fluorescence intensity of the refolded protein
4604PRASAD ET AL.APPL. ENVIRON. MICROBIOL.
was found to be low. The pH of the medium after cell growth
for 8 h was observed to be 6.03 ? 0.10 (Table 1). Since the
stability of GFP has been reported to decrease steadily below
pH 6.5 (1), the pH of the growth medium was maintained
above 7.0 by dissolving LB broth in 0.1 M phosphate buffer, pH
7.8, instead of distilled water. After growth of cells and induc-
tion of protein expression for 8 h, the pH of the medium was
measured and the cells were lysed as before. In all cases, the
volume of the soluble fraction was adjusted to that of the
finally refolded urea-solubilized fraction for ease of compari-
son. The final pH of the medium after 8 h of induction was
found to be 7.47 ? 0.05 when LB broth was dissolved in 0.1 M
phosphate buffer, pH 7.8. The expression of GFP as inclusion
bodies was ?35-fold higher when the solvent for the medium
was 0.1 M phosphate buffer, pH 7.8 (Table 1; see Fig. S1 in the
supplemental material). It has been shown in in vitro systems
that the folding of GFP was inhibited at a pH of ?6.5 (1). In
our case, we found that as the pH of the medium decreased to
below 6.5, the fluorescence of GFP was reduced. This could
not be reversed when the urea-solubilized fraction was diluted
with lysis buffer. The pH instability of GFP could be overcome
by preparing the growth medium buffered at pH 7.8. Thus, all
further experiments were carried out in LBampmedium pre-
pared in 0.1 M phosphate buffer, pH 7.8.
Effect of sorbitol on solubilization of GFP. Sorbitol was
added to the growth medium at different concentrations (0.1 M
to 0.7 M; dissolved in distilled water) along with the inducer
(0.2% [wt vol?1] arabinose), and the cells were allowed to grow
as before. The pH of the medium (prepared in distilled water)
was measured after the cells were harvested and was found to
decrease when sorbitol was present in the growth medium
(Table 1). This was presumably due to the generation of sugar
acids. When LB broth was dissolved in 0.1 M phosphate buffer,
pH 7.8, the pH of the medium increased after 8 h of induction
(Table 1). The fluorescence intensity of GFP in the soluble
fraction (cell lysate) was monitored and was ?12-fold higher
when the cells were grown in a buffered medium in the pres-
ence of 0.5 M sorbitol than when they were grown in a medium
of distilled water (Table 1; see Fig. S1 in the supplemental
material). Sorbitol was found not to interfere with the fluores-
cence emission intensity of GFP at all concentrations studied
(data not shown). As the concentration of sorbitol increased
from 0.1 M to 0.7 M, the fluorescence intensity of GFP in the
soluble fraction was found to increase until 0.5 M sorbitol
(Table 2; see Fig. S2 in the supplemental material). Beyond
this concentration, the rate of cell growth was affected drasti-
cally (as measured by A600) and the fluorescence intensity of
GFP in the soluble fraction decreased (Table 2), although no
FIG. 1. Expression of GFP. (A) Transformed E. coli BL21(DE3)
cells were plated on LBampplates containing 0.2% (wt vol?1) arabinose
(a), 0.2% (wt vol?1) arabinose and 0.2% (wt vol?1) glucose (b), 0.2%
(wt vol?1) arabinose and 0.4% (wt vol?1) glucose (c), and 0.2% (wt
vol?1) glucose (d). Results for uninduced (e) and untransformed (f)
cells plated on LBampplates (containing no supplements) are also
shown. (B) Partitioning of GFP fluorescence into soluble and insoluble
fractions. Spectra of soluble fraction (I), urea-solubilized pellet (II),
and 1:1 (vol vol?1) dilution of urea-solubilized pellet (III) are shown.
The fluorescence emission spectrum of the urea-solubilized fraction of
uninduced cells (IV) is shown for comparison. All spectra were re-
corded at an equal concentration of protein except spectrum III.
TABLE 1. Effect of pH of medium on expression of GFPa
Medium and additive pH
Medium prepared in distilled
0.4 M sorbitol
0.5 M sorbitol
6.03 ? 0.10
5.45 ? 0.07
5.23 ? 0.05
198 ? 6
881 ? 19
1,341 ? 85
Medium prepared in 0.1 M
phosphate buffer, pH 7.8
0.4 M sorbitol
0.5 M sorbitol
7.47 ? 0.05
7.23 ? 0.02
7.15 ? 0.07
6,838 ? 202
10,150 ? 327
14,514 ? 415
aCells were grown in solid LB broth dissolved in either distilled water or 0.1
M phosphate buffer, pH 7.8. Induction was carried out with 0.2 % (wt vol?1)
arabinose with and without additives. After 8 h, the cells were centrifuged and
the pH of the medium was measured. The harvested cells were lysed and ana-
lyzed for the expression of GFP. Results reported are mean values ? standard
deviations of three independent experiments. AU, arbitrary units.
TABLE 2. Effect of sorbitol on solubilization of GFPa
2.15 ? 0.03
2.09 ? 0.03
2.06 ? 0.04
1.95 ? 0.03
1.92 ? 0.03
1.89 ? 0.04
1.53 ? 0.03
1.22 ? 0.04
6,841 ? 201
6,137 ? 271
5,564 ? 156
5,922 ? 216
9,136 ? 256
10,104 ? 239
8,710 ? 257
7,997 ? 242
aInduction was carried out with 0.2 % (wt vol?1) arabinose with and without
different concentrations of sorbitol. Sorbitol was found not to interfere with the
measurement of fluorescence intensity of GFP. Results reported are mean val-
ues ? standard deviations of three independent experiments. AU, arbitrary
VOL. 77, 2011INHIBITION OF INCLUSION BODY FORMATION BY OSMOLYTES4605
concomitant increase in the urea-soluble fraction was seen
(data not shown). The growth rate of the transformed cells in
the presence of 0.5 M sorbitol was lower (0.486 h?1) than that
of cells grown in the presence of 0.2% (wt vol?1) arabinose
without any additive (0.695 h?1). The fluorescence intensity
per mg of total protein in the soluble fraction at 0.5 M sorbitol
was 1.4-fold higher than that obtained when inclusion bodies of
GFP were solubilized in 8 M urea. The expression of total
protein in the cell lysate (soluble fraction) was not significantly
affected by the presence of sorbitol in the growth medium.
Thus, the increase in the fraction of soluble GFP in the cell
lysate occurred due to the presence of sorbitol in the growth
Effect of arginine on solubilization of GFP. Different con-
centrations of arginine (0.1 to 0.25 M) were included in the
growth medium along with the inducer, and the cells were
grown as described before. After the cells were harvested, they
were lysed and the fluorescence intensity of GFP in the soluble
fraction was measured. The fluorescence emission of GFP in
the soluble fraction was found to increase with the increase in
the concentration of arginine (Table 3; see Fig. S3 in the
supplemental material). The presence of arginine at these con-
centrations did not interfere with the fluorescence emission of
GFP. At and beyond 0.2 M arginine, the cell growth was
affected (as measured by A600) (Table 3). The growth rate of
the transformed cells in the presence of 0.2 M arginine was
lower (0.550 h?1) than that of cells grown in the presence of
0.2% (wt vol?1) arabinose without any additive (0.695 h?1) but
higher than the growth rate observed in the presence of 0.5 M
sorbitol (0.486 h?1). The fluorescence intensity of GFP in the
soluble fraction increased until 0.2 M arginine and then de-
clined. The fluorescence intensity per mg of total protein in the
soluble fraction at 0.2 M arginine was 2.4-fold higher than that
obtained when inclusion bodies of GFP were solubilized in 8 M
urea. Thus, the presence of arginine led to the formation of a
larger amount of soluble GFP, and the maximum solubilization
occurred at a concentration of 0.2 M. This is much lower than
the reported concentration of arginine (1 and 2 M) required
for solubilization of GFP in vitro when it is expressed as inclu-
sion bodies (35).
Effect of ethylene glycol on solubilization of GFP. Trans-
formed E. coli BL21(DE3) cells were grown in the presence of
different concentrations of ethylene glycol (5 to 15%, vol
vol?1) along with the inducer (0.2% [wt vol?1] arabinose) as
described before. Cells were harvested and lysed, and the flu-
orescence intensity of the soluble fraction was monitored. The
fluorescence intensity of the lysate for the cells grown in the
presence of 10% (vol vol?1) ethylene glycol was maximum but
was less than that observed in the urea-solubilized fraction of
the cells grown in the presence of 0.2% (wt vol?1) arabinose
alone (Table 4; see Fig. S4 in the supplemental material). As
can be seen, the cell growth declined with increasing concen-
trations of ethylene glycol. The growth rate of the transformed
cells in the presence of 10% (vol vol?1) ethylene glycol was
much less (0.274 h?1) than the growth rates of cells grown in
the presence of 0.2% (wt vol?1) arabinose (0.695 h?1) and 0.5
M sorbitol (0.486 h?1) or 0.2 M arginine (0.550 h?1). The
presence of ethylene glycol was found not to interfere with the
fluorescence emission of GFP. Thus, ethylene glycol, a known
protein stabilizer, was not able to solubilize GFP completely.
Mechanism of solubilization of GFP. The fluorescence in-
tensity of GFP in the soluble fraction was the maximum when
the cells were grown in the presence of 0.2 M arginine (2.4-
fold), followed by 0.5 M sorbitol (1.4-fold), compared to the
fluorescence intensity of GFP in the urea-solubilized insoluble
fraction when cells were grown in the presence of the inducer
only (see Fig. S5 in the supplemental material). This could be
confirmed by the fluorescence intensity of the residual pellet
following cell lysis (see Fig. S5 in the supplemental material).
GFP has been reported to retain its functional form in the
insoluble fraction (14, 17, 35). This was established by the
fluorescence of the colonies on the LBampplate supplemented
with 0.2% (wt vol?1) arabinose (Fig. 1A). The increased re-
sidual fluorescence intensity of the pellet obtained in the case
of cells grown in the presence of the inducer alone also con-
firmed that the increased fluorescence intensity of GFP ob-
served in the presence of compatible solutes was because of
solubilization of overexpressed GFP and not due to enhanced
fluorescence of the existing soluble protein.
The amounts of sorbitol and arginine in the cell lysates were
estimated by HPLC (28) and colorimetry (29), respectively.
Neither sorbitol nor arginine could be detected in the soluble
fraction (cell lysate). The amount of osmolyte required for
solubilization of GFP was also much lower than the amount
required if stabilization were to be carried out in a manner
similar to that in in vitro situations. Neither trehalose nor
glycerol, the two osmolytes which are commonly produced in
cells when they are exposed to stress conditions, could be
detected in the transformed cells grown in the presence of
either sorbitol or arginine (see Fig. S6 in the supplemental
The ATP content of the cells grown in the presence of
TABLE 3. Effect of arginine on solubilization of GFPa
2.14 ? 0.03
2.03 ? 0.07
1.65 ? 0.06
1.10 ? 0.05
6,741 ? 213
5,598 ? 235
16,400 ? 188
14,213 ? 156
aInduction was carried out with 0.2% (wt vol?1) arabinose with and without
different concentrations of arginine. Arginine was found not to interfere with the
measurement of fluorescence intensity of GFP. Results reported are mean val-
ues ? standard deviations of three independent experiments. AU, arbitrary
TABLE 4. Effect of ethylene glycol on the solubilization of GFPa
Concn of ethylene
2.12 ? 0.04
1.98 ? 0.03
1.50 ? 0.04
0.93 ? 0.05
6,756 ? 200
2,167 ? 151
5,800 ? 211
5,612 ? 242
aInduction was carried out with 0.2% (wt vol?1) arabinose with and without
different concentrations of ethylene glycol. The additive was found not to inter-
fere with the measurement of fluorescence intensity of GFP. Results reported
are mean values ? standard deviations of three independent experiments. AU,
4606 PRASAD ET AL.APPL. ENVIRON. MICROBIOL.
sorbitol, arginine, and fructose was measured in search of a
mechanism for the above observation. The solubilization of
GFP was found to be proportional to the concentration of ATP
produced in the transformed cells grown in the presence of the
additives along with the inducer (Fig. 2A). When untrans-
formed cells and cells transformed with GFP-containing or
empty vectors were grown in the presence of 0.2% (wt vol?1)
arabinose without any additive, the ATP contents of the cells
were almost similar. However, in the presence of 0.5 M sorbitol
or 0.2 M arginine, there was a significant increase in the ATP
content of the transformed cells, and this corresponded well
with the degree of solubilization of GFP. In the presence of 0.5
M fructose, an increase in the ATP content of the untrans-
formed cells compared with that of cells grown in the presence
of the inducer alone was observed. However, no further in-
crease in the ATP content of the transformed cells grown in
the presence of 0.5 M fructose and 0.2% (wt vol?1) arabinose
(Fig. 2A) was observed. The presence of GFP was found not to
interfere with the luminescence values observed due to pro-
duction of ATP (see Fig. S7 in the supplemental material). The
cells were grown in the presence of 0.5 M fructose along with
the inducer, and the amount of soluble GFP was analyzed.
Native PAGE analysis showed that the amount of soluble GFP
was negligible (Fig. 2B). Densitometric analysis (ImageQuant;
GE Healthcare) showed that the intensity of the band was of
the same level as that formed in the presence of 0.2% (wt
vol?1) arabinose (inducer) alone, in the absence of any
osmolyte, and in the presence of 10% (vol vol?1) ethylene
glycol. The maximum amount of soluble GFP was seen in the
presence of 0.2 M arginine, as expected from the spectrofluo-
rimetry data (Table 3).
Although no additive (sorbitol or arginine) could be de-
tected in the cell lysate, only those molecules which could
permeate the membrane barrier and were potential partici-
pants in ATP generation pathways were successful in solubi-
lizing the overexpressed protein. Molecules which could cross
the barrier but were not aggregation suppressors, e.g., fructose,
were not successful in solubilizing GFP. Those molecules
which were protein stabilizers but could not cross the mem-
brane (e.g., ethylene glycol) were able to solubilize the over-
expressed protein only to a small extent. The mechanism of
stabilization is complex and is not limited only to the forces
involved in the interaction between compatible osmolytes and
proteins prevalent under in vitro conditions.
Expression and solubilization of HypF-N. The successful
solubilization of overexpressed GFP described above was with
a protein which retains its functional conformation inside in-
clusion bodies. To confirm that the observed stabilization is a
general phenomenon and that the strategy can be extended to
other proteins, we selected HypF-N as an illustrative case.
HypF codes for the hydrogenase maturation factor in E. coli.
The N-terminal domain of this protein (HypF-N) is reported
to form inclusion bodies when it is expressed in E. coli (38).
Since it is an endogenous bacterial protein, inclusion body
formation can be studied in its natural milieu.
E. coli BL21(DE3) cells were transformed with an HypF-N–
His6-containing plasmid construct. Induction of protein ex-
pression was carried out with 1 mM IPTG, and the expression
of the target protein was monitored by SDS-PAGE. Most of
the protein was found to be expressed as inclusion bodies
which could be solubilized using 8 M urea (Fig. 3A). For
expression of soluble HypF-N, transformed cells were grown in
the presence of different concentrations of sorbitol and argi-
nine. The cells were lysed after induction of protein expression
with 1 mM IPTG as described in the Materials and Methods
section. The expression level of soluble HypF-N was monitored
using anti-His6antibody. The amount of soluble HypF-N var-
ied with the concentration of osmolytes used (data not shown).
The maximum solubilization was again observed with 0.5 M
sorbitol and 0.2 M arginine (Fig. 3B). Densitometric analysis
(ImageQuant; GE Healthcare) showed that in case of the
protein expressed in the presence of inducer alone, only ?6%
of HypF-N was present in the soluble fraction (considering the
intensity of the protein band corresponding to the urea-solu-
bilized fraction to be 100%). In the presence of 0.2 M arginine
or 0.5 M sorbitol along with the inducer in the growth medium,
the fraction of soluble protein increased to 24% and 68%,
respectively. Thus, both arginine and sorbitol were able to
express HypF-N in the soluble form. We repeated the exper-
iment in the presence of 0.5 M trehalose, a molecule which has
FIG. 2. Mechanism of solubilization of GFP in E. coli BL21(DE3) cells. (A) ATP content of cells which were untransformed (gray bars),
transformed with GFP (black bars), and empty vectors (white bars) and grown in the presence of 0.2% (wt vol?1) arabinose and 0.5 M sorbitol,
0.2 M arginine, or 0.5 M fructose. Experiments were carried out in triplicate, and the mean and standard error of three independent experiments
are shown. (B) Native PAGE of soluble fractions obtained after cell lysis. An equal amount of protein (20 ?g each) was loaded in the wells of a
12% cross-linked polyacrylamide gel. After the electrophoretic run, the band intensity of GFP was visualized with an image scanner (Typhoon Trio;
GE Healthcare). Lane 1, uninduced cells; lane 2, induced cells; lane 3, induced cells in the presence of 0.5 M sorbitol; lane 4, induced cells in the
presence of 10% (vol vol?1) ethylene glycol; lane 5, induced cells in the presence of 0.5 M fructose; lane 6, induced cells in the presence of 0.2
VOL. 77, 2011INHIBITION OF INCLUSION BODY FORMATION BY OSMOLYTES4607
found in widespread use as a protein stabilizer (18, 20). Tre-
halose is also a potential participant in glycolysis via its con-
version to glucose and further to glucose-6-phosphate (3). In
the presence of 0.5 M trehalose, HypF-N was found to be
expressed in the soluble form in E. coli. The extent of solubi-
lization was 58% compared with the band intensity of urea-
solubilized HypF-N expressed in the presence of inducer
alone. The amount of solubilized protein formed in the case of
HypF-N was different and less than that formed in the case of
GFP compared to the amount for the corresponding urea-
solubilized inclusion bodies. It is possible that since GFP is
already in the functional form in the inclusion bodies, the
efficiency of refolding is much higher than that in the case of
HypF-N, where the protein is in a crossed ?-sheet conforma-
tion. The amount of trehalose in the cell lysate was estimated
and was found to be 14.4 ? 0.5 mM. This confirms that only
those osmolytes which are able to enter the cell and are prob-
able participants in its energy generation pathways are effective
solubilizers of overexpressed proteins.
Over the years, it has been shown that proteins present in
inclusion bodies share many common characteristics with ag-
gregated proteins found under other conditions, like aggrega-
tion of proteins during manufacture and storage or in protein-
misfolding diseases. Thus, any strategy developed to prevent
the formation of inclusion bodies in cells can be extended, with
appropriate modifications, to other scenarios of protein aggre-
gation. We employed commonly used protein stabilizers for
expression of soluble proteins in E. coli and obtained counter-
intuitive results with some of them. Sorbitol is used quite
commonly as a stabilizer of proteins in vitro (13, 21). It is able
to inhibit the unfolding of native conformations to the unfolded/
misfolded forms by a mechanism similar to that of other poly-
hydric alcohols (20, 39). Sorbitol can be transported across E.
coli via the sorbitol-specific phosphoenolpyruvate phospho-
transferase system in the form of sorbitol-6-phosphate (22, 23).
Sorbitol-6-phosphate can enter glycolysis by the action of sor-
bitol-6-phosphate dehydrogenase, which converts it to fruc-
tose-6-phosphate (19, 30), a key intermediate of glycolysis.
Similarly, the role of arginine as an aggregation suppressor of
proteins has been widely recognized (2, 26, 35). The effect of
arginine on proteins has been compared with that of guani-
dinium hydrochloride (GdHCl), inasmuch as both of them
interact favorably with the side chains of proteins (2). How-
ever, unlike GdHCl, arginine interacts only with the residues
on the surface of the protein, thus acting as a stabilizer. Argi-
nine has been included in the solubilization buffer for inclusion
bodies wherein chaotrope-induced denaturation of the over-
expressed protein was not required (35). Transport of arginine
into E. coli occurs via two routes. At higher concentrations, the
more specific, low-affinity system, consisting of an arginine-
binding protein and a membrane transporter, operates (7, 8).
Inside the cell, arginine gets converted to succinate by the
successive action of at least five different enzyme systems (34)
and can enter the tricarboxylic acid (TCA) cycle.
Another osmolyte that has been used for stabilization of
proteins against aggregation, though not as frequently as sac-
charides or arginine, is ethylene glycol (15, 16). The inability of
ethylene glycol to act as a denaturing agent for proteins is
attributed to its ability to increase the free energy of contact
between the nonpolar side chains and solvent (water) and its
inability to decrease the free energy of hydrophobic interac-
tions between them (27). The fact that solutes which can par-
ticipate in energy-yielding pathways are the ones which act as
efficient solubilizers of proteins in cells is evident from the fact
that ethylene glycol, which has no known transporter in wild-
type E. coli (4), is not able to solubilize GFP completely. In all
cases, solubilization occurs until the osmolyte concentration
overwhelms cell growth. As the cell growth declines, so does
the expression of soluble fluorescent GFP.
In order to understand if the presence of an intermediate of
the glycolysis/TCA cycle is enough to solubilize the overex-
pressed protein, transformed cells were grown in the presence
of 0.5 M fructose, a monosaccharide which has no reported
activity as an aggregation suppressor. The entry of fructose
into E. coli under high-fructose conditions is facilitated by a
low-affinity phosphotransferase system that converts fructose
to fructose-6-phosphate (11). Under these conditions, GFP
was not expressed as a soluble protein. We have shown that
both GFP and HypF-N overexpressed in E. coli were present as
soluble proteins when protein expression was carried out in the
presence of compatible solutes. These molecules have earlier
been shown to act as protein stabilizers in vitro. This property
alone, however, was not enough to prevent the proteins from
forming inclusion bodies. As seen in the case of ethylene gly-
col, which has no known protein destabilization effect, the
observed solubilization of GFP was quite low. We found that
the metabolic state of the cell could be directly correlated with
the extent of solubilization of the overexpressed protein. In the
case of protein stabilizers which increased the level of ATP
produced by the cells, the protein did not form inclusion bodies
and was expressed in the soluble form. However, mere involve-
FIG. 3. Expression of His6–HypF-N. (A) SDS-PAGE of the sam-
ples was carried out using a 15% cross-linked polyacrylamide gel. The
gel was Coomassie stained. The arrow indicates the expected position
of HypF-N-His6. (B) Western blotting of His6–HypF-N in the soluble
fraction after growth of cells in the presence of different osmolytes.
Lane M, molecular mass marker proteins; lane 1, soluble fraction (cell
lysate), uninduced cells; lane 2, urea-solubilized fraction, induced cells;
lane 3, soluble fraction (cell lysate), induced cells; lane 4, soluble
fraction (cell lysate), cells induced in the presence of 0.5 M trehalose
and inducer; lane 5, soluble fraction (cell lysate), cells induced in the
presence of 0.5 M sorbitol and inducer; lane 6, soluble fraction (cell
lysate), cells induced in the presence of 0.2 M arginine and inducer.
4608 PRASAD ET AL.APPL. ENVIRON. MICROBIOL.
ment in the energy-generating pathway of the cell was not
enough. Only compatible osmolytes, i.e., molecules which
could interact favorably with protein side chains and stabilize
them against inactivation and which could potentially contrib-
ute to ATP generation by the cell, proved to be effective solu-
bilizers of the overexpressed proteins and inhibited formation
of inclusion bodies.
Conclusion. Solubilization of overexpressed proteins in E.
coli could be achieved by including osmolytes in the medium
during protein expression. Those osmolytes which could per-
meate the cell membrane and contribute to the energy gener-
ation pathways of the cell were able to solubilize the recombi-
nant proteins. Those molecules which either could not cross
the membrane barrier or were unable to participate in the
energy-yielding processes were incapable of increasing the sol-
ubility of the target proteins. Judicious choice of additives
during protein expression can thus increase the solubility of
recombinant proteins in E. coli and can inhibit the formation
of inclusion bodies.
We are extremely grateful to Jonathan Weissman (University of
California, San Francisco, CA) for the gift of pBAD.GFPwt599 and to
Ronald Riek (ETH, Zurich, Switzerland) for the gift of HypF-N-His6
constructs. We acknowledge helpful discussions with Lei Wang (Salk
Institute, CA) regarding the primary growth conditions of HypF-N and
Vikas Grover (Central Instrumentation Laboratory, NIPER) for car-
rying out HPLC analysis. We are thankful to Ankan Kumar Bhadra for
critical reading of the manuscript.
1. Alkaabi, K. M., A. Yafea, and S. S. Ashraf. 2005. Effect of pH on thermal and
chemical-induced denaturation of GFP. Appl. Biochem. Biotechnol. 126:
2. Arakawa, T., et al. 2007. Suppression of protein interactions by arginine: a
proposed mechanism of the arginine effects. Biophys. Chem. 127:1–8.
3. Boos, W., et al. 1990. Trehalose transport and metabolism in Escherichia coli.
J. Bacteriol. 172:3450–3461.
4. Boronat, A., E. Caballero, and J. Aguilar. 1983. Experimental evolution of a
metabolic pathway for ethylene glycol utilization by Escherichia coli. J. Bac-
5. Bradford, M. M. 1976. A rapid and sensitive method for the quantitation of
microgram quantities of protein utilizing the principle of protein-dye bind-
ing. Anal. Biochem. 72:248–254.
6. Carrio ´, M. M., R. Cubarsi, and A. Villaverde. 2000. Fine architecture of
bacterial inclusion bodies. FEBS Lett. 471:7–11.
7. Celis, T. F. R., H. J. Rosenfeld, and W. K. Maas. 1973. Mutant of Escherichia
coli K-12 defective in the transport of basic amino acids. J. Bacteriol. 116:
8. Celis, T. F. R. 1981. Chain-terminating mutants affecting a periplasmic bind-
ing protein involved in the active transport of arginine and ornithine in
Escherichia coli. J. Biol. Chem. 256:773–779.
9. Crameri, A., E. A. Whitehorn, E. Tate, and W. P. Stemmer. 1996. Improved
green fluorescent protein by molecular evolution using DNA shuffling. Nat.
10. Curtis-Fisk, J., R. M. Spencer, and D. P. Weliky. 2008. Native conformation
at specific residues in recombinant inclusion body protein in whole cells
determined with solid-state NMR spectroscopy. J. Am. Chem. Soc. 130:
11. Ferenci, T., and H. L. Kornberg. 1974. The role of phosphotransferase-
mediated syntheses of fructose 1-phosphate and fructose 6-phosphate in the
growth of Escherichia coli on fructose. Proc. R. Soc. Lond. B Biol. Sci.
12. Ferreira, J. C., V. M. F. Paschoalin, A. D. Panek, and L. C. Trugo. 1997.
Comparison of three different methods for trehalose determination in yeast
extracts. Food Chem. 60:251–254.
13. Filatova, L. Y., S. C. Becker, D. M. Donovan, A. K. Gladilin, and N. L.
Klyachko. 2010. LysK, the enzyme lysing Staphylococcus aureus cells: specific
kinetic features and approaches towards stabilization. Biochimie 92:507–513.
14. Garcia-Fruitos, E., et al. 2005. Aggregation of bacterial inclusion bodies
does not imply inactivation of enzymes and fluorescent proteins. Microb.
Cell. Fact. 4:27.
15. Gekko, K., and T. Morikawa. 1981. Thermodynamics of polyol-induced
thermal stabilization of chymotrypsinogen. J. Biochem. 90:51–60.
16. Gekko, K., X. Li, and S. Makino. 1999. Competing effects of polyols on the
thermal stability and gelation of soy protein. Biosci. Biotechnol. Biochem.
17. Gonza ´lez-Montalba ´n, N., E. García-Fruito ´s, and A. Villaverde. 2007. Re-
combinant protein solubility—does more mean better? Nat. Biotechnol.
18. Jain, N. K., and I. Roy. 2009. Effect of trehalose on protein structure. Protein
19. Jones-Mortimer, M. C., and H. L. Kornberg. 1976. Uptake of fructose by the
sorbitol phosphotransferase of Escherichia coli K12. J. Gen. Microbiol. 96:
20. Kaushik, J. K., and R. Bhat. 2003. Why is trehalose an exceptional protein
stabilizer? An analysis of the thermal stability of proteins in the presence of
the compatible osmolyte trehalose. J. Biol. Chem. 278:26458–26465.
21. Kumar, V., V. K. Sharma, and D. S. Kalonia. 2009. Effect of polyols on
polyethylene glycol (PEG)-induced precipitation of proteins: impact on sol-
ubility, stability and conformation. Int. J. Pharm. 366:38–43.
22. Lengeler, J. 1975. Mutations affecting transport of the hexitols D-mannitol,
D-glucitol, and galactitol in Escherichia coli K-12: isolation and mapping. J.
23. Lengeler, J., and E. C. C. Lin. 1972. Reversal of the mannitol-sorbitol diauxie
in Escherichia coli. J. Bacteriol. 112:840–848.
24. Li, M., Z. G. Su, and J. C. Janson. 2004. In vitro protein refolding by
chromatographic procedures. Protein Expr. Purif. 33:1–10.
25. Lillie, S. H., and J. R. Pringle. 1980. Reserve carbohydrate metabolism in
Saccharomyces cerevisiae: responses to nutrient limitation. J. Bacteriol. 143:
26. Liu, Y. D., et al. 2007. A newly proposed mechanism for arginine-assisted
protein refolding—not inhibiting soluble oligomers although promoting a
correct structure. Protein Expr. Purif. 51:235–242.
27. Nozaki, Y., and C. Tanford. 1965. The solubility of amino acids and related
compounds in aqueous ethylene glycol solutions. J. Biol. Chem. 240:3568–
28. Papesa, S., D. Jezek, and D. Kujundzic. 2001. Determination of sorbitol
concentration in diet chocolate by high-performance liquid chromatography.
Food Technol. Biotechnol. 39:129–133.
29. Rosenberg, H., A. H. Ennor, and J. F. Morrison. 1956. The estimation of
arginine. Biochem. J. 63:153–159.
30. Roux, C., L. Salmon, and C. Verchere-Beaur. 2006. Preliminary studies on
the inhibition of D-sorbitol-6-phosphate 2-dehydrogenase from Escherichia
coli with substrate analogues. J. Enzyme Inhib. Med. Chem. 21:187–192.
31. Sambrook, J., and D. W. Russell. 2001. Molecular cloning: a laboratory
manual, 3rd ed. Cold Spring Harbor Laboratory Press, Cold Spring Har-
32. Schein, C. H., and M. H. M. Noteborn. 1988. Formation of soluble recom-
binant proteins in Escherichia coli is favoured by lower growth temperature.
Biotechnology (NY) 6:291–294.
33. Schleif, R. 2000. Regulation of the L-arabinose operon of Escherichia coli.
Trends Genet. 16:559–565.
34. Schneider, B. L., A. K. Kiupakis, and L. J. Reitzer. 1998. Arginine catabo-
lism and the arginine succinyltransferase pathway in Escherichia coli. J.
35. Tsumoto, K., R. Abe, D. Ejima, and T. Arakawa. 2010. Non-denaturating
solubilization of inclusion bodies. Curr. Pharm. Biotechnol. 11:309–312.
36. Wang, C., and Y. Cheng. 2010. Urea-gradient protein refolding in size ex-
clusion chromatography. Curr. Pharm. Biotechnol. 11:289–292.
37. Wang, L., S. K. Maji, M. R. Sawaya, D. Eisenberg, and R. Riek. 2008.
Bacterial inclusion bodies contain amyloid-like structure. PLoS Biol. 6:e195.
38. Winkelmann, J., et al. 2010. Low-level expression of a folding-incompetent
protein in Escherichia coli: search for the molecular determinants of protein
aggregation in vivo. J. Mol. Biol. 398:600–613.
39. Wu, P., and D. W. Bolen. 2006. Osmolyte-induced protein folding free energy
changes. Proteins 63:290–296.
40. Wu, B., et al. 2011. Expression, refolding and purification of a human inter-
leukin-17A variant. Cytokine 53:107–114.
VOL. 77, 2011 INHIBITION OF INCLUSION BODY FORMATION BY OSMOLYTES 4609