Gene, 99 (1991) 121-126

121

Elsevier

GENE

03949

Adsorption

to starch of a /I-galactosidase

fusion protein containing

the starch-binding

region of

Aspergillus glucoamylase (Recombinant

DNA;

enzyme)

Luojing Chen”, Clark Ford a and Zivko Nikolov b ‘IDepartment of Genetics and ‘Department Received by J.A. Gorman: 10 August Revised: 24 October 1990 Accepted: 25 October 1990

of Food Technology, Iowa State University, Ames, IA 50011 (U.S.A.)

Tel. (515)294-3157

1990

SUMMARY

We have constructed and purified by affinity chromatography three B-galactosidase @Gal) fusion proteins (BSB133, BSBCD8, and BGA134) containing amino acid (aa) sequences from Aspergillus glucoamylase (GA). BSB133, containing the C-terminal 133 aa of GA (aa 484-616) adhered to native starch granules with a much higher affinity (Kad = 18 ml/g starch) than a PGal control (K,, = 0.9 ml/g starch). Two other fusion proteins, BSBCD8 and BGA134, similar in size to BSB133, adhered to starch with a relatively low affinity (K,, = 7 ml/g starch, and K,, = 4 ml/g starch, respectively). BSBCD8 differs from BSB 133 by a truncation of 8 aa at the C terminus. BGA134 contains 134 aa from an overlapping region of GA (aa 380-513). These results confirm the presence of a strong starch-binding region (SBR) included in the C-terminal 133 aa of GA and indicate that the SBR can confer starch-binding activity on a fusion protein produced in Escherichia coli. In the presence of crude soluble cell extracts, the fusion proteins adsorbed by native starch granules with an affinity similar to that of the purified enzymes. BSB 133 that had been adsorbed by starch from crude extracts could be eluted at a high level of purity, similar to that achieved by affinity chromatography. These results suggest that it may be feasible to use native starch as an adsorbent for the recovery and purification of recombinant fusion proteins containing the SBR. Starch has many favorable qualities for this application: it is inexpensive, stable, nontoxic, and easy to recover by centrifugation.

INTRODUCTION

Genetically engineered polypeptide ‘tails’ have been used by several groups to promote recovery and purification of recombinant fusion proteins by affinity chromatography

Correspondence to: Dr. C. Ford, University,

Department

Ames, IA 50011 (U.S.A.)

of Genetics,

Iowa

State

Tel (515)294-0343;

Fax (5 15) 294-0907. Abbreviations:

aa, amino acid(s); Ap, ampicillin;

bp, base pair(s); adsorption otide(s);

GA, glucoamylase;

constant PAGE,

(ml/g starch);

LB, Luria kb, kilobase

polyacrylamide-gel

/?Gal, /I-galactosidase; Bertani

electrophoresis;

binding region; SDS, sodium dodecyl sulfate; wt, wild type; (fusion).

0378-I

119/91/$03.50

0

1991

Elsevier

Saence

Publishers

(broth);

K,,,

or 1000 bp; nt, nucle-

B.V

SBR,

starch-

: : , novel joint

(reviewed by Sassenfeld, 1990). We are working to develop a lower-cost system of fusion-protein recovery applicable to industrial use that utilizes the SBR of Aspergillus GA as a fusion ‘tail’ and native starch granules as adsorbent. Starch is plentiful and inexpensive, stable, non-toxic, and easy to recover in batch or continuous processes by sedimentation or centrifugation. Recently, two other systems involving the adsorption of fusion proteins to complex carbohydrates have been reported (Ong et al., 1989; Bedouelle and Duplay, 1988). GA from Aspergillus sp. (EC 3.2.1.3) exists in two forms: GA1 (aa 1-616) has a large catalytic domain (aa l-440) a Ser/Thr-rich highly O-glycosylated region (aa 441-512), and a C-terminal domain (aa 513-616), containing a strong SBR, which allows the enzyme to be adsorbed by (and

122 therefore digest) native starch granules (Svensson et al., 1982; 1986a); GA11 (aa 1-512) in the proteolytic product of GAI, lacks the C-terminal domain, and adheres to native starch granules with low affinity (K,, = 1.5 ml/g starch) in comparison to GA1 (Kad = 49 ml/g starch; Dalmia, 1990). The location of a strong SBR in the C-terminal domain is consistent with the observation of significant aa sequence homology between this domain in GA and starch-binding domains of other carbohydrases (Svensson et al., 1989). Hayashida et al. (1989) have reported some starch-binding activity from a GA peptide fragment of 45 aa (470-514) located primarily in the 0-glycosylated region. Since this peptide is almost completely included in GA11 we believe its starch-binding to be of minor importance. Native starch has successfully been used as an adsorbent to purify GA1 from GA11 on the basis of the SBR (Ueda et al., 1974). Starch has a high binding capacity for GA1 (4.9 mg protein/g starch; Dalmia, 1990) and the bound enzyme can be eluted under gentle conditions with a yield of > 90% (Medda et al., 1982; Dalmia, 1990). These results suggest that fusion proteins containing the GA1 SBR could be recovered and purified by adsorption to and elution from native starch granules. In this paper we report the construction of and adsorption by starch granules of a pGa1 fusion protein (BSB 133) containing the C-terminal 133 aa of A. awamori GA. Our results confirm the presence of a strong SBR at the C terminus of GA and indicate that the SBR is active when expressed intracellularly in E. cofi.

EXPERIMENTAL

AND DISCUSSION

(a) Construction of fusion proteins The E. coli plasmid vector pUR290 was used for the construction and expression of three plasmids (pLC1, pLC2, and pLC3) that encode /IGal fusion proteins containing aa sequences from A. awumori GA (Fig. 1). The fusion proteins contain full-length flGa1 with the GA residues at the C terminus. BSB133 contains 133 aa from the C terminus of GA (aa 484-616). This fusion therefore includes the 104 aa of the C-terminal domain essential for strong GA1 starch-binding activity. BSBCD8 differs from BSB133 by the truncation of 8 aa at the GA C terminus. The truncated region includes Trp615 suggested by Svensson et al. (1986b) to be important for starch binding based on chemical modification studies. BGA134 is a control fusion protein (about the same size as BSB133) which contains 134 aa from GA (aa 380-5 13), and overlaps 29 aa of BSB133 (aa484-513). BGA134 includes 44 ofthe aa (470-514) in the 0-glycosylated region identified by Hayashida et al. (1989) to have some starch-binding activity.

(b) Synthesis

and purification

of fusion proteins

The vectors pLC1, pLC2, pLC3, and pUR290, encoding BSB133, BSBCD8, BGA134, and BGal, respectively (Fig. l), were transformed into a protease-deficient E. coli Ion- strain for synthesis of the fusion proteins and a BGal control. The proteins were purified by affinity chromatography and analyzed by SDS-PAGE (Fig. 2). All of the fusion proteins exhibited proteolytic degradation to a similar extent resulting in two major bands: a band of about 130 kDa representing full-length fusion protein monomers, and a band of about 116 kDa, approximately the size of a wt BGal monomer. Similar degradation of fiGa fusion proteins has been observed by others (Ruther and Mtiller-Hill, 1983; Germino and Bastia, 1984; Hellebust et al., 1988). Immunoblots (not shown) verified that both bands react to anti-/IGal antibody. The specific activities of the fusion proteins (Table I) were nearly as high as that of /IGal, suggesting that the GA SBR does not interfere greatly with BGal function. The specific activities in Table I, however, may be an overestimate of the true specific activities of the fusion proteins since the majority of the purified enzyme tested consisted of 116 kDa monomers, and the possibility exists that these monomers (which resulted from proteolysis) were more active than the full-length fusion protein monomers. (c) Adsorption by starch To determine the relative starch-binding ability of the purified fusion proteins, a starch-binding assay was performed (Fig. 3a). BSB133 was adsorbed by starch with an affinity 18-fold higher than that of the PGal control, as shown by K,, values. This indicates the presence of a strong SBR included in the C-terminal 133 aa of GA. Although the affinity to starch of BSB 133 was much higher than that of fiGal, it was approx. threefold lower than that reported by Dalmia (1990) for GA1 (Kad = 49 ml/g starch), measured under slightly different conditions. This was not surprising since the size and conformation of PGal fusion proteins are different from those of GAL In addition, proteolytic degradation of the fusion proteins likely results in an underestimation of their true binding affinity. The higher affinity of BSB 133 to starch compared to that of pGal was not due simply to the presence of additional residues at the C terminus of flGa1 as shown by the much lower affinity of the similar-sized fusion proteins BGA134 and BSBCD8. The reduced affinity to starch of BSBCD8 compared to BSB133 indicated that deletion of 8 aa from the C terminus of BSB 133 sharply decreased the function of the SBR. This is consistent with the results obtained by chemical modification of GA Trp615 (Svensson et al., 1986b) that indicated this residue may be essential for starch binding. The lower affinity to starch of BGA134 compared to BSB133 was expected based on the dif-

123

a . ..SSIVDAVKTF ADGPVSIVET 371 380

HAASNGSMSE

QYDKSDGEQL

SARDLTWSYA

ALLTANNRRN

SWPASWGET

SASSVPGTCA 441

ATSAIGTYSS

VTVTSWPSIV

ATGG~TTAT

PTGSGSVTST 484

SKTTATASKT

STSTSSTSCT

TPTAVAVTFD 513

LTATTTYGEN

IYLVGSISQL

GDWETSDGIA

LSADKYTSSD

PLWYVTVTLP

AGESFEYKFI

RIESDDSVEW 581

ESDPNREYTV

PQACGTSTAT 608

VTDTWR

b

c

PLCl

. ..CAA AAA GGG GAT CGA TCC GGC... . ..Gln LVS Gly Asp Arg Ser Gly... 1023 484

PLCZ

. ..CAA AAA GGG GAT CGA TCC GGC... TCG ACT AGA CTA GTC TAG,.. . . .Gin Lvs Gly Asp Arcj Ber Gly . ..Ser Thr Arg Leu Val --1023 484 608

pIlc3

. ..CAA AAA GGG GAT CCT TTC GCC...CCC . ..Glnlz. Gly Asp Pro Phe Ala...Pro 380

BmtHI HindIII . ..CAA AAA GGG GAT CCG TCG ACC TGC AGC CAA GCT TAT CGA TGA... . ..GLnlg. Gly ASP Pro Ser Thr Cys Ser Gln Ala Tyr Arg ---

puri290

Fig. 1. Construction from Nunberg

of fusion proteins

Standard

and plasmids.

et al. (1984). (b) ,!IGal fusion proteins

The vector pUR290 methods

(R&her

and wolfer-~iil,

for DNA manipulation

the lucZ gene for the construction BumHI

+ IlindIIl-cut

pGAC9

was a generous

was constructed

(a)C-terminril238

containing

aa offf, awumori GA (aa 371-616)

GA ‘tails’ at the C terminus.

1983) was used as a starting were used (Sambrook

of C-terminal

fusions

containing

full-length

fragment

from pGAC9

gift of Cetus Corporation. The resultant

by inserting

The reading

plasmid,

a SpeI linker containing

pUR290

stop codons

in one-letter

indicate

code. Sequence

of gene fusions en~oding~Gai::GA pLC0

(not

(Innis et al., 1985) encoding of pLC0 was corrected

the fusion protein BSB133. Plasmid

shown)

was

(Evans fragment protein

containing

GA residues

et al., 1990) encoding

380-513.

GA residues

from pRE5 13 contains

at aa 513. The nt sequences Nucleic Acid Facility.

pLC3 was constructed

at the junction homologous

The universal

by ligating pUR290

The XmnI site of the fragment

part of the GA-encoding

(5’-GCGGAATTCCAGCTGAG) University

Plasmid 380-513.

gene from pGAC9

with the addition

by tilling-in the cohesive

of the gene fusions were verified by the dideoxy

to the DNA sequence

site of pUR290 method

the nt sequence

BSBCD8,

Biolabs) into the

Sal1

site

a control fusion protein fragment

were blunted

linker which truncates

(Chen and Seeburg,

from pRE513

before ligation.

The

the encoded

1985) using a primer

near the 3’ end of the la&? gene. The primer was synthesized

Ml3 primer was used to determine

ends at the

the fusion protein

New England

of a SpeI stop-codon

by ligating

133 aa of GA. Plasmid

with a 0.85 kb XmnI-liindII1

and the BarnHI

protein fusions.

constructed

the C-terminal

near the 3’ end of the GA coding region of pLC1. The .Su(Isite was biunted before ligation with the linker. Plasmid pLC3 encodes (BGAI34)

is

constructions.

and Hind111 sites at the 3’ end of

pLC2, which encodes

in three frames (5’~CTAGACTAGTCTAG,

information

aa from GA. (c) Plasmid

has unique BarnHI

j3Gal. Plasmid

frame at the fusion junction

pLC1, encodes

Numbers

point for the construction

et al., 1989). Plasmid

pUR290 with a 0.55-kb BarnHI-Hind111

&rplrHI site and religating.

ACC TAG... Thr --513

at the Iowa State

at the 3’ end of the pLC2 gene fusion. Sequence

illformation for pUR290 is from Rtither and Miller-Hill (1983). In part c, the underlined aa are from the authentic C terminus ofFGal and are numbered according to their position in wt BGal. Bold-face aa are from GA, and are numbered according to their position in wt GA. For pLC2, aa following GA residue

608 are encoded

by the +eI

linker.

affinity to starch of GA1 compared to GAIL Adsorption of the fusion proteins to starch in the presence of crude soluble enzyme preparations was assayed to

ferential

determine the effect of contaminating background proteins on the specificity of starch-protein interaction (Fig. 3b). To make accurate comparisons, equal amounts of each puri-

124

1

2

3

4 -205

Fig. 2. SDS-PAGE

of purified fusion proteins.

pLC2, pLC3 and pUR290 were expressed

The IacZ genes of pLCl,

in the protease-deficient

Ion host strain Y 1089-l and the flGal fusion proteins were purified

by affinity chromatography.

from the strain Y1089 {dlacU169,

-116 -97

grown

(encoding

/IGal)

during phase,

following matrix

(Sambrook

p-aminophenyl

/&r+thiogalactopyranoside-agarose method

Coomassie Lanes:

v 0

PGal

gel containing

brilliant blue dye according

1, BSB133;

2, BSBCD8;

to starch

in crude extracts was assayed

The data from duplicate

of purified

fusion proteins

(b). Adsorption modified

indicate

the apparent

acetate

washing

buffer

buffer)

volume

constants

pH 6.0. (a) Affmity-purified

were added

with initial protein

to washed concentrations

(unbound)

at

(K.,