Vol.

176,

May

15, 1991

No.

3, 1991

BIOCHEMICAL

AND

BIOPHYSICAL

RESEARCH

COMMUNICATIONS

Pages

1462-1468

PROTEINSEPARATIONAND PURIFICATION IN NEAT DINETHYLSULFOXIDE

Nancy Chang,

Department

Received

April

J. Hen, and Alexander N. Kli banov*

Stewart

of Chemistry, Cambridge,

8,

Massachusetts Massachusetts

Institute 02139

of Technology

1991

SUMMARY: Pure DMSO (instead of water) is used as the reaction medium for protein separations. It is shown that common extracellular proteins (i) have high solubility in DMSO (l-50 mg/ml), (ii) do not irreversibly inactivate in this solvent, and (iii) can adsorb onto carboxymethyl cellulose in DMSO and be subsequently fully desorbed in this solvent by inorganic salts. Ion-exchange chromatography on this resin in DMSO has been used to purify bovine pancreatic trypsin and to separate it from hen egg-white lysozyme in their mixture. Another approach to protein separation in DMSO, fractional precipitation with ethyl acetate (which does not dissolve proteins), has been verified with a mixture of bovine pancreatic chymotrypsinogen and chicken egg ovalbumin. 0 1991 AcademicPress, Inc. The advent

of modern

protein

separation

efficient bottleneck proteins

can contribute methods

be easily

scaled

recoveries

(5).

new approaches

systems

metal

filtration,

partition No single

an optimal upon

the

In this

options.

study,

is based on the

(DMSO) , instead

Abbreviations:

using

should

DMSO,dimethyl

(7),

separation

and affinity

into

by

cross-flow

(10).

each

we explore as the

always

and insufficient research

case.

of a diverse

use of the

of water,

' To whom correspondence

for

cannot

extraction

(8,9),

of

Traditional

(2-4) of recent

methodology

availability

(1).

e.g.,

is

scheme comprised

has to be selected

purification

a surge

micelles

for

a major

has been

and reversed

purification

contingent

is

processing,

and precipitation

bioseparation

the

costs

economics

need

processing

and purification

chromatography

techniques

which

there

(6)

affinity protein

processing

of unattractive protein

ever-growing

manufacturing:

separation

Consequently,

immobilized

thus

up to 90% of the

to downstream

to the

Downstream

protein

of protein

up because

two-phase

has led

methodologies.

in recombinant-DNA-based

bench-scale

aqueous

biotechnology

a novel neat

of range approach

ideal,

a combination

The success

non-aqueous

process

universally

of

of this

of protein

specific strategy

separation

to protein solvent

and

dimethyl

separations sulfoxide

medium.

be addressed. sulfoxide;

0006-291X/91 $1.50 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

CM-cellulose,carboxymethyl

1462

cellulose.

is

Vol.

176,

No.

BIOCHEMICAL

3, 1991

EXPERIMENTAL

AND

BIOPHYSICAL

RESEARCH

COMMUNICATIONS

PROCEDURES

Materials. Enzymes and other proteins were purchased from Sigma Chemical Co.: hen egg-white lysozyme (EC 3.2.1.17) and ovalbumin; and bovine pancreatic trypsin (EC 3.4.21.4), chymotrypsin (EC 3.4.21.1), ribonuclease (EC 3.1.27.5), were obtained and cr-chymotrypsinogen A. All chemicals and solvents DMSO (Aldrich Chemical commercially and were of analytical grade or purer. Co.) was 99.9% pure and used without further purification. Assays. Protein determinations were carried out according to the Lowry method (11) following at least a l&fold dilution of a DMSO solution with water (10 Calibration curves were obtained beforehand for mM phosphate buffer, pH 7.0). each protein under each set of conditions. Lysozyme and trypsin activities were determined spectrophotometrically using dried cells of Micrococcus lvsodeikticus (12) and N-a-p-toluenesulfonylL-arginine methyl ester (13), respectively, as substrates. The concentration of the catalytically competent active centers in trypsin was measured in water by spectrophotometric titration with R-nitrophenyl R'-guanidinobenzoate (14). The individual concentrations of chymotrypsinogen and ovalbumin in their mixtures were determined by HPLC. Protein solutions in DMSO were diluted lOOfold with 8 M aqueous solutions of urea and then concentrated 5-fold by ultrafiltration using Amicon microconcentrators (Centricon 10). The resultant samples were injected onto a Waters Protein Pak (300 SW) HPLC size exclusion co1 umn. The mobile phase was 10 mM aqueous phosphate buffer, pH 7.5 (flow rate of 1 ml/min); the protein content was monitored by absorbance at 280 nm. To determine their solubility in DMSO, proteins were first dissolved in distilled water at 10 mg/ml, the pH was adjusted to the desired value, and the solutions were lyophilized. Each solid protein sample was then added to DMSO and stirred at 30°C. Following complete dissolution, more protein was added, and the procedure was repeated until no more protein dissolved. The undissolved protein was subsequently removed by centrifugation, and the protein concentration in the supernatant was measured by the Lowry method.

RESULTS AND DISCUSSION Although there

are

a few

significant that

it

proteins

are

solvents,

such

concentrations may be feasible

dissolved

in such

has not

been

protein

conformations,

in almost

to apply

classical

instead

before.

non-aqueous that

(15).

be different

dissolve

techniques

to our

non-covalent

solvents,

Consequently,

separation

of water;

Since

should

all

as DMSO and formamide,

of common proteins

solvents

studied

insoluble

to proteins

knowledge,

this

interactions,

in organic

we reasoned approach

and hence

media

compared

to water

behavior in them should be distinct as well, (I5)Y separation expanding the overall scope of bioseparation opportunities.

In addition,

working

with

practical

obstacle

in

be removed

such because

In this review, solvent, water, dissolves

solvents

protein

enzymes

are

investigation,

such as high of odor most

catalytically

we chose

organic

stability, and color,

of water,

degradation

low

inactive

and many inorganic 1463

in this

of attractive

toxicity,

and relatively

a major

by endogenous

DMSO to test

18 and 19) has a number

see refs. lack

as DMSO instead

bioseparations,

infinite low

compounds

cost.

thereby

proteases,

such media

by will

(15-17).

approach.

DMSO (for

properties

as a

miscibility

with

This

polar

and can be easily

solvent

a

Vol.

176,

No.

separated

from

has found

a score

First, model

water

enzymes,

concentrations

be prepared,

far

conformation

matter

enzymes

whether following

lyophilization. with

enzymes

water.

after

a 24-hour

enzymes

studied,

Having we then is

have

medium

(23),

lysozyme from to

should With

CM-cellulose,

question

we initially

solubility

in this

was added enzyme

by inorganic (2-4).

solvent

with

it salts,

this

was found adsorbed

The same result

to other and bovine

was restored buffer

but

at least

in

isoelectric

We tested

protein

and the per

is possible as routinely

was obtained

adsorption lyophilized resin)

was

an hour.

to desorb done

the

in salts

lithium moles/liter. complete with

of

10 ml of DMSO, up

gram of the

simplest,

to be several

at pH 6.0,

resin in

point

in a non-aqueous the

many inorganic

on CM-cellulose,

in neat DMSO,

solution

at 30°C within

1464

or

technique

charge

suspended

one of the

it

enzymes

aqueous

enzyme

Out of the

selected

to lysozyme

occurred.

from

on CM-cellulose

is

water

was inactive

Thus

lysozyme's

retaining

200 mg of the

activity

aqueous

The first

Since

the

not

inactivation.

medium.

ion-exchanger

was whether

in water

DMSO (19),

the

(i.e.,

adsorbed e.g.,

chromatography

in DMSO (both

does

with

activity with

to cation-exchangers.

25 mg of the

enzyme

completely The next

LiCl

adsorb

in water it

chymotrypsin,

of working

and,

that

by simple

in DMSO at 25°C.

lyophilized

charge

to CM-cellulose

5 mg of the

found

enzyme,

positive

the

and extended

dilution

in this

because

trypsin

be recovered

reversible

feasibility

tested

enzymatic

while

enzymatic

chromatography.

the

a net

pH 6.0).

original

enzymes

upon dilution

lysozyme,

immediate

separation

from

as the

observation

of enzymes

the

their

11 (22),

around

upon

incubation

was ion-exchange

should

all

be easily

standpoint

that

could

including

DMSO causes only

established

explored

examined

only

different

reported

this

pH 6.0)

All

DMSO (15-17)

e.g.,

activity

could

same conditions.

a bioseparations

(16)

Nearly not

in neat

procedure,

solutions

pancreatic

solvent.

is grossly

We confirmed

ribonuclease.

the

in DMSO as long

separation

from

DMSO solutions

the

active

under

Two

into

bovine

peroxidase

in this

from

100% of its

(lyophilized

pancreatic even

are

that

Rees and Singer

in DMSO, essentially dilution

stability

DMSO.

in DMSO at

solidified

including

10 mg/ml

solvent

note

the

gradually

inactive

in this

However,

restored

enzyme

DMSO

a-chymotrypsin,

dissolved

and horseradish

been catalytically

(15, 20, 21).

they

at 1 to

neat

even more concentrated

proteins,

egg ovalbumin,

we examined

have

protein

from

standing

in

pancreatic

readily

(3O'C);

Consequently,

solubility

and bovine

both

other

in DMSO at least

Next, thus

upon

Several

chicken

dissolved

lysozyme

COMMUNICATIONS

applications.

of protein

as 50 mg/ml

but

gels.

trypsin,

issue

at pH 6.0,

as high

transparent

the

RESEARCH

lyophilized.

and therapeutic

hen egg-white water

BIOPHYSICAL

and also

of industrial

from

AND

by distillation

we addressed

lyophilized could

BIOCHEMICAL

3, 1991

enzyme

from

ion-exchange soluble chloride,

in whose

When 0.5 desorption

1.0 M NH,NO,.

M of

Thus

it

Vol.

176,

No.

BIOCHEMICAL

3, 1991

AND

BIOPHYSICAL

RESEARCH

COMMUNICATIONS

0.0

Volume,

ml

Fiwre 1. CM-Cellulose column chromatography of hen egg-white lysozyme (A), bovine pancreatic trypsin (B), and their mixture (C) in neat DMSO. A glass column (13x2 cm) was packed with 9 g of granular CM-cellulose lyophilized from water (containing 20 mM citrate - 60 mM phosphate buffer) at pH 6.0. The packed column was equilibrated with DMSO at 3O'C and the flow rate of 1.2 ml/min. Then the column was loaded with 15 mg of either enzyme (or 7.5 mg each in the case of their mixture) dissolved in DMSO at 5 mg/ml (both enzymes had been lyophilized from water at pH 6.0). Following washing of the column with DMSO, the enzymes were eluted with LiCl solutions in DMSO at the salt concentrations indicated above the arrows in the figure. Throughout the chromatography, 3.5-ml fractions were collected and assayed for protein and enzymatic activity as described in Experimental Procedures. For lysozyme, the unit of enzymatic activity is a 0.1 decrease in absorbance at 450 nm per minute at 25'C; for trypsin the international units are given (13).

is

possible

it

when

to desired,

Figure

single recovery

pancreatic

neat

adsorbed and 96%

the

the

that

experiments trypsin.

results

the

were Active

site

that

that

peak.

ion-exchanger

a column

see

and

an

in

chromatography of

can

column

for

on

for

One

activity and

enzyme

allowing

DMSO.

to

protein was

the

thereby

in

Similar

adsorb

1A depicts

CM-cellulose applied

both

In

a typical

out

titration 1465

this

entire

amount

elution

desorb

of

lysozyme

of

enzyme

yielded

experiment, activity

and

solvent.

chromatography the

subsequent

enzymatic

carried

in

DMSO

the

the

on

same

protein

91%.

with

another

(14)

of

Sigma's

enzyme,

bovine

commercial

sample

Vol.

176,

No.

of the

enzyme

dissolved lB),

BIOCHEMICAL

revealed

only

64% of the contained

0.05

M LiCl,

In agreement increase

to

exchange

protein

and the

found

can be separated achieved,

as illustrated

trypsin

went

by 0.05

M LiCl;

While

whereas

the

ion-exchange aqueous

aqueous

solution

trypsin

bound

following

binding

0.25

M to be eluted.

was measured

capacity

is

capacity

of CM-cellulose

resin

not

resin

which from

to be the

same)

in organic We then

separations solvents

increases

in DMSO.

If

were

of the

when the resin

and,

lyophilized,

in the

amount

lyophilized solution

to 10.0, in the

raised.

as it

is

yet proteins.

independent

another, out

binding

ionization the

pH of the (the

two do

operational in enzymatic

earlier, the vast majority Hence when such a solvent

in DMSO, at some point

dependence

of protein 1466

solubility

the

protein

which

These

enzyme

approach

the from

from

the

Thus the

in

of enzyme The

were

may be a useful media,

was suspended

respectively.

presumably,

trypsin

dependence).

changes

pH is

both

(23).

As pointed

solution the

mg/ml,

the

present;

aqueous

or raised

lB),

amount

both

to 2.0

two peaks

(i.e.,

total

to 0.32

in non-aqueous

dissolve

to a protein

the

peak

of this when

pH of the

in terms

solvents

investigated

do not

precipitate.

when the

increased

which

in bioseparations

catalysis

added

or

into (Fig.

and the

of the

point

by the ptl of

For example,

was observed

was added,

lysozyme

was lowered

to 0.10

solution

variable

however,

affected

and inactive

behavior

onset

for

can be rationalized

of the have

as the

was lyophilized

differences state

mg/ml;

dropped

aqueous

defined

eluted

salt

in DMSO seems the

enzyme

as a function

of

and was then

(25 mg of CM-cellulose

lysozyme

binding

the

Similar

lyophilized

fraction

was chromatographed active

a single

experiments

binding

pH 6.0 was 0.24

pH 6.0)

and trypsin

was indeed

profoundly

was lyophilized.

resin

pH 6.0,

column).

is

the

from

capacity

supernatant

in water

ion-

DMSO due to their

at that

parameter

from

lysozyme

adsorbed

adsorbed

was

Thus

enzyme.

inactive

analogous

pH 3.0 yielded

to the

10 ml of DMSO, then

trypsin

chromatography

from

that

to

purity

of trypsin

the

the

its

40%.

in neat

remained

(lyophilized

activity

Such a separation

active

(Fig.

was eluted

showed

purifies

however,

lyophilized

fractions

specific

column,

which

fraction

by approximately

chromatography

was

The non-adsorbed

1A and 1B suggest

the

sample

titration

1C: as expected,

from

lyophilized

the

ion-exchanger.

solution,

on CM-cellulose

capacity

result,

site

COMMUNICATIONS

chromatography

column.

by Fig.

lysozyme,

When this

The adsorbed

chromatography

to this

and required

pH of the

to the active

in Figs.

RESEARCH

column

in DMSO significantly

through

concentration

was 57%.

activity.

this

the

BIOPHYSICAL

to CM-cellulose

by CM-cellulose

affinities

in the

purity

competent

presented

different

AND

adsorbed

with

after

chromatography

The data

resin

its

no enzymatic

be 80%.

the

that

in DMSO and subjected

fraction with

3, 1991

to protein is

of organic gradually

should

in DMSO on the

Vol.

176,

No.

BIOCHEMICAL

3, 1991

AND

BIOPHYSICAL

RESEARCH

COMMUNICATIONS

0-l 0

Concentration

acetate, %(v/v)

of ethyl

20

40

Concentration

60

acetate, %(v/v)

of ethyl

of ovalbumin (curves a) and chymotrypsinogen (curves Fioure 2. Precipitation h) from their solutions in DMSO, induced by ethyl acetate. The proteins were lyophilized from pH 6.0 (A) or pH 9.1 (B). For other conditions, see text.

concentration

of the

different

proteins, This

acetate l),

Figure pH 6.0)

from

ethyl

acetate

the

An even two

b),

this

more marked

proteins

contrast

was observed

Comparison

also

less

the

when they

of curves may be used

to separate

bioseparation

was carried

out

both

(lyophilized

it

proteins and stirred

formed

was separated

DMSO and, content

the

along

with

of both

the

the

those

2B indicates

precipitate

supernatant,

precipitate

independently;

further

from

from

2 suggests

As seen for

in

ovalbumin.

behavior pH 9.1

that

from

was

DMSO by ethyl than

precipitation

lyophilized

ethyl

of the

(Fig.

28).

acetate

chymotrypsinogen.

1 ml of DMSO containing

9 mg/ml

Such a each

of

We added 0.8 ml of ethyl acetate 30 min at 3O'C. The precipitate

for

the pellet was analyzed

the

1467

was also

of ethyl

Comparison proteins are

investigation

to

was redissolved in neat by HPLC; the protein

supernatant

was found that the addition but only 32% of ovalbumin. when

of

at higher

pH 6.0).

was needed

ovalbumin

and the

that

(lyophilized

at 50% of ethyl

from

fractional

pH 9.1).

mixture

by centrifugation,

It

not

from

resultant

Lowry assay. of chymotrypsin in Fig.

with

Curve

The same experiment

was precipitated

were

in DMSO

30% (v/v)

but

of solution;

lyophilized

a and b in Fig.

precipitation

of ovalbumin

transparent,

acetate

ethyl

chymotrypsinogen.

profile

out

ethyl

in

solvent

by chromatography

precipitated.

(also

protein

significantly

separation.

Up to approximately

to fall

for

protein

common organic

remained

protein

chymotrypsinogen

2A (curve but

the

and distinct

for

pancreatic

acetate.

began

all

sharp

separated

precipitation

solution

ovalbumin with

acetate,

the

protein

the

those

and bovine

DMSO by ethyl

essentially

conducted

than

is

can be used

using

other

2A represents

concentrations

phenomenon

egg ovalbumin

from

co-solvent

was verified

chicken

acetate,

this

proteins

a in

Fig.

then

rationale

and two

(Fig.

non-dissolving

acetate

determined

by the

precipitated

69%

of these data with present together they of this

phenomenon

is

do

Vol.

176,

No.

underway.

At any rate,

solvent

fractional

we have

proteins

strategy

our

in DMSO.

to other as well

separation

in non-aqueous

proteins

suitable where

it

ACKNOWLEDGMENTS. Engineering Center

AND

BIOPHYSICAL

validate

the

demonstrated Future

that

work

will organic

as to understanding for

solvents. the

may obviate

the

the

if it

is

it

solvents mechanistic

used

separation

toward

by

repeatedly.

to separate expanding

and this

and separation basis

described

of membrane

COMMUNICATIONS

of protein is

possible

be directed

The approach

purification

RESEARCH

idea

in particular

protein-dissolving

techniques particularly

results

precipitation,

In closing, purify

BIOCHEMICAL

3, 1991

of protein herein

and other

may be hydrophobic

use of detergents.

This research was supported by NSF's at MIT. S.J.H. is an NSF predoctoral

Biotechnology fellow.

Process

REFERENCES

:: 3. 4. 5.

Dwyer, J.L. (1984) Bio/Technol. 2, 957-964. Scopes, R.K. (1987) Protein Purification: Princioles and Practice, 2nd edn., Springer-Verlag, New York. Burgess, R.R. (1987) Protein Purification, Alan R. Liss Press, New York. Harris, E.L.V. & Angal, S., eds. (1989) Protein Purification Methods: A Practical Aoproach, IRL Press, Oxford. Lillehoi. E.P. & Malik, V.S. (1989) Advan. Biochem. Enq./Biotechnol. 40,

19-71. “’ 6. 7. 8. 9. 10. 11. 12. :34: 2 17.

Hustedt, H. Kroner, K.H., Menge, U. & Kula, M.R. (1985) Trends Biotechnol. 3, 139-144. Dekker, M., Hilhorst, R. & Laane, C. (1989) Anal. Biochem. 178, 217-226. Porath, J..( 1988) Trends Anal. Chem. 7, 254-259. Arnold, F.H (1991) Bio/Technol. 9, 151-156. Luong, J.H. T *, Nguyen, A.L. & Male, K.B. (1987) Trends Biotechnol. 5, 281-286. Lowry, O.H. Rosebrough, N.J., Farr, A.L. & Randall, R.J. (1951) JBiol. Chem. 193, 265-275. Shugar, D. (1952) Biochim. BioDhvs. Acta 8, 302-309. Hummel, B.C .W. (1959) Can. J. Biochem. 37, 1393-1399. Hruska, J.F., Law, J.H. & Kezdy, F.J. (1969) Biochem. Bioohvs. Res. Commun. 36, 272-277. Singer, S.J. (1962) Advan. Protein Chem. 17, l-68. Rees, E .D. & Singer, S.J. (1956) Arch. Biochem. Bioohvs. 63, 144-159‘. Klyosov , A.A., Van Viet, N. & Berezin, I.V. (1975) Eur. J. Biochem. 59, - 3-l.

18.

19. 20.

Jacob, S.W., Rosenbaum, E.E. & Wood, D.C., Eds. (1971) Dimethvl Sulfoxide, M. Dekker, New York. Gaylord Chemical Corp. (1985) Dimethvl Sulfoxide (DMSO) Technical Bulletin, Slidell, LA. Clore, G.M., Martin, S.R. & Gronenborn, A.M. (1986) J. Mol. Biol.

1911,

553-561. 21. 22. 23.

Hofmann, M., Gondol, D., Bovermann, G. & Nilges, M. (1989) Eur. J. Biochem. 186, 95-103. Imoto, T., Johnson, L.N., North, A.C.T., Phillips, D.C. & Rupley, J.A. (1972) in The Enzymes, 3rd edn. (Boyer, P.D., Ed.), vol. 7, pp. 665-868, Academic Press, New York. Klibanov, A.M. (1989) Trends Biochem. Sci. 14, 141-144.

1468