[ 16] REFOLDING OF RECOMBINANT PROTEINS 187

the limits of tolerance are for substituting protease moieties with recombi-
nant protein sequences. We also need to develop a better understanding of
the hemolysin system since this pathway may become a means of secreting
proteins that cannot be secreted by other pathways. The motivation of the
genetic engineer to achieve secretion of heterologous proteins in E. coli
ensures that progress will be made.

Acknowledgments
The authors would like to thank P. Bassford, P. C. Tai, D. Oliver, and B. Holland for
communicatingdata prior to publication and T. Hirst, T. Pugsley, and M. Better for helpful
discussions. The authors would also like to thank S. Benson and M. Igo for critically reading
the manuscript and S. DiRenzo for typing.

[ 16] R e f o l d i n g o f R e c o m b i n a n t Proteins
B y TADAHIKO KOHNO, DAVID F. CARMICHAEL, ANDREAS SOMMER, and
ROBERT C. THOMPSON

Proteins that are normally made in the cells of higher organisms are
now routinely made in microorganisms by cloning the gene together with
appropriate signals for transcription and translation. However, the protein
made is often inactive because the polypeptide chain does not readily
adopt its native conformation in these organisms. This result is not to be
taken as evidence against the view that the primary structure of a protein
determines its conformation. More reasonably, it indicates that the protein
will only fold to its correct conformation under the appropriate conditions,
and that these conditions are not met in the microorganism. The failure to
fold properly may even induce the protein to form a precipitate, com-
monly known as an inclusion body, within the microorganism.
To induce the recombinant protein to assume its active conformation,
two broad strategies have evolved. The most elegant, and intellectually
pleasing, strategy is to recreate in the microorganism the conditions pre-
vailing in the higher cell. Most frequently, this corresponds to cotransla-
tionally secreting the protein to an oxidizing environment, either the peri-
plasmic space or the extracellular medium. This strategy is remarkably
successful for some proteins.L2 However, when some eukaryotic proteins
are designed to secrete cotranslationally in microorganisms, the proteins
appear to poison the microorganism. 2 For this reason, an additional

t j. A. Stader and T. J. Silhavy, this volume [ ! 5].

: T. Kohno et al., manuscript in preparation.

Copyright 1990 by Academic Press, Inc.

METHODS IN ENZYMOLOGY, VOL. 185 All rights of reproduction in any form reserved.
188 EXPRESSION IN E. coli [ 16]

strategy has been employed in which the protein is first made in a disor-
dered state in the microorganism and is subsequently folded to its native
state in vitro. This strategy, and the methods employed, are derived almost
exclusively from the large volume of work on the in vitro refolding of
denatured proteins that was initiated by Anfinsen and collaborators in the
early 1960s. a However, few reports describe the refolding of recombinantly
produced proteins. In this article, we illustrate this strategy by describing
examples of folding two highly disulfide-bonded human proteins from the
disordered state present in Escherichia coll.
The process of refolding of the recombinant protein, to be feasible for
commercial application, should be fast and cheap and give a high yield of
an active molecule. The strategy used employs three essential steps. In the
first of these, the insoluble protein aggregate from the microorganism is
effectively solubilized and partially purified. In the second, the solubilized
protein is brought into an environment which favors the native, active
structure. Finally, since it is too much to expect the folding to proceed in
100% yield, it is necessary to find some method to remove the residual,
wrongly folded protein.

Solubilization of Inactive Proteins

The solubilization of insoluble aggregates can be achieved by various
denaturing reagents. Low or high pH can solubilize some inclusion bodies,
presumably by increasing electrostatic repulsions between protein mole-
cules, thereby destabilizing aggregates.4 High temperature, detergents, and
high concentration of inorganic salts or organic solvents will also solubilize
some aggregates) Some proteins, such as DNase I, can be refolded after
treatment with a detergent SDS. 6 However, in some cases, detergents may
bind strongly to the protein, 7 inhibiting subsequent refolding. The most
commonly used solubilizing agents are high concentrations of water-solu-
ble organic solutes such as urea or guanidine-HC1. These are often used in
the presence of reducing agents like mercaptoethanol or dithiothreitol
(DTT) to break aggregates that are stabilized by disulfide bonds. The
solubilized proteins are often partially purified by ion-exchange chroma-

3 C. J. Epstein, R. F. Goldberger, and C. B. Anfinsen, Cold Spring Harbor Symp. Quant.

Biol. 28, 439 (1963).
4 j. S. Emtage, S. Angal, M. T. Docl, T. J. R. Harris, B. Jenkins, G. Lilley, and P. A. Lowe,
Proc. Natl. Acad. Sci. U.S.A. 80, 3671 (1983).
5 I. Bikel, T. M. Roberts, M. T. Bladon, R. Green, E. Amann, and D. M. Livingston, Proc.
Natl. Acad. Sci. U.S.A. 80, 906 (1982).
6 S. A. Lacks and S. S. Springhorn, J. Biol. Chem. 255, 7467 (1980).
7 C. Tanford, Adv. Protein Chem. 23, 121 (1968).
[16] REFOLDING OF RECOMBINANT PROTEINS 189

tography or other conventional methods at this stage, prior to the refolding

process. For some proteins it is not absolutely required that the protein be
purified prior to the refolding; however, recovery of the active molecule
seems to be higher when this is done.

Refolding of Proteins
In general, the refolding of proteins is initiated by complete unfolding
of the protein of interest. Very often this will have been accomplished by
the conditions needed to dissolve the aggregates. However, if this is not the
case, strong denaturants, such as urea or guanidine-HC1, and reducing
agents are added at this stage. If no disulfide bonds are present in the final
conformation, it would probably be sufficient to remove the denaturing
agent slowly to allow the protein to fold correctly. However, when disulfide
bonds are to be formed, it is frequently beneficial to form them duringthe
course of refolding since they stabilize the native structure and thereby
influence the process.
The formation of disulfide bonds during the refolding of a protein can
be done by an air oxidation, s including oxidation by air catalyzed by the
presence of trace metal ions, 9 or by the presence of a mixture of reduced
and oxidized thiol compounds, 9 or by disulfide isomerase) Air oxidation
is a slow process. However, the addition of the thiol-disulfide exchanger
increases the rate of reactivation. Refolding buffers containing conjugate
thiol:disulfide pairs such as oxidized DTT and reduced DTT, oxidized
glutathione (GSSG), and glutathione (GSH), cystine and cysteine, or cys-
tamine and cysteamine yield similar results in some cases. 9 The appropri-
ate ratio of oxidized and reduced thiols will vary and must be determined
experimentally.
The formation of protein aggregates during the refolding is sometimes
observed and will reduce the recovery of the active molecule. This aggrega-
tion may be limited by the presence of a low concentration of urea or
guanidine-HC1 during the refolding. But even when this is true, it is
sometimes necessary to use very low concentrations of protein (-10 gg/ml)
to avoid the formation of insoluble aggregates. The pH of the solution may
affect protein aggregation as well as the rate of disulfide interchange.
Optimal pH of buffer must therefore be determined empirically to maxi-
mize the recovery of an active molecule. The removal of denaturants used
s C. B. Anfinsen, E. Haber, M. Sela, and F. H. White, Proc. NatL Acad. Sci. U.S.A. 47, 1309
(1961).
9 p. Saxena and D. B. Wetlaufer, Biochemistry 9, 5015 (1970).
,o D. F. Carmichael, J. E. Morin, and J. E. Dixon, J. BioL Chem. 252, 7163 (1977).
190 EXPRESSION IN E . cold [16]

can be done either by dialysis or by dilution in an appropriate buffer. The

strategy we have adopted is first to oxidize all the free --SH groups to
intermolecular disulfide bonds with a suitable low-molecular-weight thiol
and then to remove the denaturing agent, relying on the fact that the
favorable energy of folding will drive the formation of intramolecular
disulfide bonds. A small amount of free thiol is incorporated into the
reaction to catalyze disulfide bond interchange.

Isolation and Characterization of Correctly Folded (Active) Proteins

Once the proteins are folded in biologically active conformation, the
active molecules have to be separated from incorrectly folded molecules.
Affinity chromatography is one of the best methods for removing inactive
molecules.
The biological activity of the refolded proteins should be identical to
that of native proteins. Purity of the refolded protein can often be assessed
by the combination of the following techniques: (1) sodium dodecyl
sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), (2) reversed-
phase chromatography (RP-HPLC), and (3) ion-exchange chromatogra-
phy (FPLC). However, it is generally not possible to show that all possible
refolding isomers will separate from the authentic protein. Besides examin-
ing the N-terminal amino acid sequence of the refolded proteins, confor-
mational identity between the recombinant proteins and human-derived
proteins could be determined by peptide mapping of those proteins with-
out prior reduction.

Example 1. Refolding of Secretory Leukocyte Protease Inhibitor

Secretory leukocyte protease inhibitor (SLPI) is a protease inhibitor
originally isolated from human tissues. SLPI inhibits various serine pro-
teases, which include leukocyte elastase, cathepsin G, and trypsin. This
inhibitor is secreted protein and has been found in various secretory fluids
such as those from the parotid glands. 1~It consists of 107 amino acids and
contains 16 cysteine residues, ll all of which are present in disulfide bonds
in the native protein. The eDNA and genomic sequence of the gene for
SLPI are available, n and the gene can be expressed in high yield in pro-
karyotic systems such as E. coil 13 However, the protein produced in E. coli
is inactive, probably due to the incorrect conformation. In order to obtain
functional SLPI from E. coli, we learned to solubilize and refold the
inactive molecules.
n R. C. Thompson and K. Ohlsson, Proc. Natl. Acad. Sci. U.S.A. 83, 6692 (1986).
~2G. L. Stetler, M. T. Brewer, and R. C. Thompson, Nucleic Acids Res. 14, 7883 (1986).
~3S. Eisenberg et al., manuscript in preparation.
[ 16] REFOLDING OF RECOMBINANTPROTEINS 191

Procedure
Initial Purification. An E. coli strain which produces recombinant SLPI
(rSLPI) at about 2% of cell protein was grown in a 10-liter fermenter at 37".
Growth was monitored by optical density. At 10 OD, SLPI synthesis was
induced by adding 0.5 m M IPTG. After 5 hr of induction, the cells were
harvested at 4 and were resuspended in 2 X volume of 50 m M Tris-HC1,
pH 7.5. The cell suspension was processed through a Gaulin mill two times
and centrifuged at 900 g for 60 min. The resultant pellet was solubilized at
25 with 50 m M Tris-HCl, pH 7.5, containing 10 M urea and 50 m M
2-mercaptoethanol. Insoluble cell debris was removed by centrifugation at
14,000 g for 60 min. The supernatant was loaded onto an SPC25 column
( 18 X 25 cm) which was previously equilibrated with 50 m M Tris-HCl, pH
7.5. The column was washed extensively with the same buffer to remove
unbound proteins. The rSLPI was eluted with 50 m M Tris-HC1, pH 7.5,
buffer containing 0.3 M guanidine-HCl. This column step yielded greater
than 80% homogeneous SLPI preparation. This eluate was concentrated to
between 0.5 and 50 mg per milliliter by ultraffltration prior to refolding.
Refolding. One liter of this concentrated SLPI was mixed with 1 liter of
6 M guanidine-HC1 and incubated at room temperature for 30 min. After
the end of 30 min, 1.23 g of solid DTT was added and gently stirred. The
incubation was continued for 1 hr at 25 , then 114 ml of 250 mMcystine
solution (in 0.5 N NaOH) was added to the above mixture. After a 10-min
incubation, this mixture was added over a period of 5 min to 10 liters of
50 m M Tris solution containing 5.3 m M cysteine. This diluted material
was incubated for up to 16 hr at 25 . The antichymotrypsin activity and
antitrypsin activity of SLPI appeared rapidly when assayed by the method
described by Thompson and Ohlsson, I~ and the refolding was almost
complete after 2 hr of incubation (Fig. 1).

o~" 100 a ~100 b

"0 "0

x x ~ x

/
0
o
or

2
50 x
o
o
rr

.;
50

/

x
/

I I I I I I I ,~ N I I I I I I I
'~: 1 2 3 4 5 6 7 1 2 3 4 5 6 7

Time (hours) Time (hours)

FIG. 1. Refoldingof rSLPI. Refoldingof rSLPI was carried out as describedin the text. At
the indicated time after refoldingstarted, samplesweredrawn and assayedfor (a) antichymo-
trypsin and (b) antitrypsin activities.
192 EXPRESSION IN E. coli [ 16]

Final Purification. After refolding, the material was purified on an

affinity column. The affinity resin is composed of anhydrochymotrypsin
bound to Affi-Gel 10 (Bio-Rad Laboratories, Richmond, CA). Anhydro-
chymotrypsin was prepared by the method described by Ako et al. ~4and
was coupled to the Affi-Gel l0 using the manufacturer's protocol. The
above refolding mixture was loaded onto the affinity column (5 X 18 cm)
and washed extensively with 50 m M sodium phosphate buffer, pH 8.0.
Active SLPI was eluted by 50 m M sodium phosphate-HC1, pH 2.0. Judg-
ing by HPLC (C8) chromatography (Fig. 2) and SDS-PAGE, the low-pH
eluate contains over 90%pure SLPI. The major contaminant appears to be
a dimer. The recovery of the pure, active SLPI by this refolding method
was about 50%, regardless of the initial protein concentration used. Evi-
dence supporting the identity of refolded rSLPI and parotid-derived SLPI
was obtained by measuring their Kd values for human leukocyte elastase.
The recombinant SLPI has a Kd of 1.9 10~M a n d that for parotid SLPI
is 2 10 ~M.
Escherichia col#produced rSLPI can be refolded to form an active
molecule without an initial purification. However, the final yield of the
active molecule was about 5 to 10%, based on a Western blot technique.

Example 2. Refolding of Recombinant Tissue Inhibitor of

Metalloproteases (rTIMP)
TIMP is an inhibitor of many mammalian metalloproteases. It is a
single polypeptide chain of 184 amino acids in length. In addition, the
primary translation product has a 23-amino acid N-terminal extension
signaling its secretory destination, t5 The mature inhibitor has 12 cysteine
residues, all of which are apparently engaged in intrachain disulfide link-
ages.t6 Although TIMP is normally glycosylated, this is not required for its
activity.~7

Procedure
Initial Purification. The 184-amino acid polypeptide chain can be pro-
duced in E. coli as an inactive, insoluble protein found predominantly in
cytoplasmic inclusion bodies containing approximately 50% pure TIMP.
In order to purify the protein further, the inclusion bodies were solub'flized
in 8 M urea, 50 m M Tris-HC1, pH 7.5, containing 0.1 M 2-mercaptoeth-
~4A. Ako, R. J. Fester, and C. A. Ryan, Biochemistry 13, 132 (1974).
~5D. F. Carmiehael, A. Sommer, R. C. Thompson, D. C. Anderson, C. G. Smith, H. G.
Welgus, and G. P. Strieldin, Proc. Natl. Acad. Sci. U.S.A. 83, 2407 (1986).
16G. P. Stricklin and H. G. Welgus, J. Biol. Chem. 258, 12252 (1983).
tv G. Stricklin, unpublished.
[ 16] REFOLDING OF RECOMBINANT PROTEINS 193

1.6

1.4

1.2

1.0 100
t$) 90

< 0.8 80

70 ~
0.6 60 ~
50 ~e -
o
0.4 40 *."
o
30 "~
0.2 20

2O 40 60 80

Time [min]
FIG. 2. HPLC pattern of affinity column eluate. A sample of low-pH eluate from the
anhydrochymotrypsin-Affi-Gel 10 column was loaded on an HPLC column (Synchrom).
The rSLPI was eluted with a 0.1% trifluoroacetic acid- acetonitrile gradient (0-50% acetoni-
trile over 50 min) at a flow rate of 1 ml/min.

anol. The insoluble material was removed by centrifugation at 30,000 g for

5 hr. The supernatant fraction was applied to a carboxymethyl-cellulose
column (Whatman CM-23) in this same buffer. Four grams of soluble
protein was loaded onto a 7.0 18 cm column. The column was washed
in 50 m M Tris-HC1, pH 7.5, containing 50 m M 2-mercaptoethanol and
finally eluted by including 0.25 M NaC1 in the wash buffer. Fractions were
analyzed for the presence of TIMP by Western blot techniques. TIMP-
containing fractions were pooled, then used in refolding reactions.
Refolding. Refolding of the partially purified protein was effected by the
stepwise addition of reagents and dilution of the denaturant from the
protein solution. All additions were made at 0 - 4 . First, 2-mercaptoeth-
anol was added to a final concentration of 140 m M i n 6.0 M urea, 50 m M
Tfis-HC1, pH 7.5. The protein was further diluted in this solution to a final
concentration of 1 mg/ml as determined by Bio-Rad protein assay kit.
194 EXPRESSION IN E. coil [ 16]

Four volumes of 250 m M cystamine in 6.0 M urea, 50 m M Tris-HCl, pH

7.5, were then added to the protein solution. Finally, the urea was diluted
to 0.3 M with the gradual addition of 50 m M Tris-HC1, pH 9.0. The pH of
the final solution was approximately 8.3. This solution was allowed to
stand at 4 overnight.
Final Purification. Following an overnight incubation, the solution was
then concentrated 20-fold by pressurized ultrafiltration on an Amicon
YM 10 filter. The buffer solution was then exchanged either by dialysis or
diafiltration for 50 m M Tris-HCl, pH 7.5. The protein at this pH was then
purified by cation-exchange chromatography from contaminating proteins
left in the solution and also from certain inactive conformers of the inhibi-
tor protein. The chromatography was performed using Whatman CM-52
carboxymethyl-cellulose. One gram of protein in solution was applied to a
2.5 15 cm column and the column was washed with 50 m M Tris-HC1,
pH 7.5. Adsorbed proteins were eluted from the column with a linear
gradient from 0 to 200 m M NaC1 in 50 m M Tris-HC1, pH 7.5. Two peaks
of inhibitor protein are partially resolved under these conditions (Fig. 3).
Fractions from the smaller peak contain inactive TIMP as well as some
degradation products. The main peak contains the active inhibitor. Pooled
fractions from this peak provide active inhibitor with a 40% recovery
through the refolding and subsequent steps.

1.0
200

8 0.5 100
E
v
<

/:,
,o 2'0 3'0 4'0 6b ;o
Fraction number
FIG. 3. CM-52 chromatogram of the refolding mixture. Both peaks contain full-length
TIMP, indistinguishableby SDS-PAGE, but only the major peak contains activeinhibitor.
[16] REFOLDING OF RECOMBINANT PROTEINS 195

Purity of the TIMP as determined by SDS-PAGE analysis is greater

than 90%, and the activity has been shown to be equivalent on a molar
basis to the inhibitor produced by human fibroblasts. In addition, the
protein produced in prokaryotic cells behaves identically to the natural
inhibitor in terms of its stability to extremes of pH, proteases, and thermal
denaturation. 17