Vol.

178,

July

31, 1991

No.

2, 1991

BIOCHEMICAL

AND

BIOPHYSICAL

RESEARCH

COMMUNICATIONS

625-633

Pages

Purification

and Characterization

of Human Recombinant Insulin-like

Growth

Factor Binding Protein 3 Expressed in Chinese Hamster Ovaq cells Timothy J. Tressel, Gwen P. Tatsuno, Kaye Spratt, and Andreas Sommerl BioGrowth, Received

May

Inc., 3065 Richmond

28,

Parkway, Suite 117, Richmond,

CA 94806

1991

Recombinant human insulin-like growth factor binding protein 3 (hIGFBP-3) stably expressed in Chinese hamster ovary cells (CHO cells) has been purified to homogeneity from serum-free culture media. The purified protein migrates as a doublet (45/43 kDa) upon SDS-PAGE. The purified recombinant hIGFBP3 is fully active and binds one mole of IGF-I per mole of recombinant binding protein. When the transfected CHO cells are treated with tunicamycin a single 29 kDa hIGFBP-3 protein is observed. This expressed hIGFBP3 protein maintains its ability to bind IGF-I. N-Glycanase treatment of the purified hIGFBP-3 protein results in a protein that migrates similar to E. coli-derived IGFBP-3 upon SDS-PAGEunder reducing conditions (30 kDa). Carboxymethylation of hIGFBP3 suggests that all 18 cysteines are involved in disulfide linkages. These results represent the first purification and characterization of recombinant hIGFBP3 expressed in CHO cells. B 1991

Academic

Press,

Inc.

Insulin-like

growth factors (IGFs) , also known as somatomedins,

are a family of

structurally related polypeptide mitogens (for review see 1 and 2). At least three classes of IGF-specific binding proteins (IGFBP-1, IGFBP-2, IGFBP3) have been characterized (3). In human serum much of the IGFs (IGF-I and IGF-II) hormone-dependent

circulate as components of a growth

“150 kDa complex” (for review see 4). This complex consists of three

components, an .acid labile protein subunit, the IGF binding protein IGFBP3 and either IGF-I or IGF-II (5). The biological role of IGFBP3 and the “15OkDa complex” is not clear. The IGFBP3 protein from human plasma has been purified and shown by SDS-PAGE to consist of two closely co-migrating protein species (6). One mole of IGFBP3 can bind one mole of IGF

(6).

The human

gene for IGFBP3

has been cloned (7, 8) and the

corresponding cDNAs predict a mature translation product of 264 amino acids including 18 cysteines, three potential

N-glycosylation

sites and two potential

O-linked

sites.

‘Person to whom correspondence should be addressed. The abbreviations used are: IGF, insulin-like growth factor; hIGFBP-3, human insulinlike growth factor binding protein-3; CHO, Chinese hamster ovary; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline; RP-HPLC, reverse-phase high pressure liquid chromatography; TFA, trifluoroacetic acid.

625

00@6-291X/91 $1.50 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.

Vol.

178,

No.

BIOCHEMICAL

2, 1991

AND

BIOPHYSICAL

RESEARCH

In this report, we describe for the first time the purification human recombinant

IGFBP-3

COMMUNICATIONS

and characterization

of

expressed and secreted by stably transfected Chinese hamster

ovary (CHO) cells. The availability of substantial quantities of glycosylated recombinant IGFBP3 should greatly facilitate further structural and functional studies of this important modulator MATERIALS

of IGF action. AND METHODS

Heparin-agarose was purchased from Sigma. The phenyl-Superose HR lO/lO column was obtained from Pharmacia. The C-4 reverse-phase semi prep and analytical HPLC columns were from Vydac. N-Glycanase was from Genzyme. ‘=I-IGF-I (2OOOCi/mmol) and iodo(2-3H)acetic acid (175 mCi/mmol) were obtained from Amersham Corp. Anti hIGFBP3 antibodies were raised in rabbits using active recombinant hIGFBP3 produced in E. coli as immunogen. E. coli-derived IGFBP3 was supplied by Rino Lee”. All other materials were the highest grade commercially available. SDS PAGE. Ligand Binding As v. and Western-Blot Assav for hIGFBP3 Proteins were electropho:sed on 10% or 15% polyacrylamide gels (9). For ligand binding assay or western blot assay, proteins were electrophoresed as described above, electroblotted to nitrocellulose (10) and probed with ‘ZI-IGF-I or hIGFBP-3 antiserum. hIGFBP-3 was visualized by autoradiography or by the Protoblot Western blot alkaline phosphatase system (Promega) respectively. Protein Concentration Determination Protein concentrations were determined using a modified Bradford assay (11,12). Bovine plasma gamma globulin (BioRad) was used as a protein standard. hIGFBP-3 concentration in serum-free culture media was determined by RIA (13). Amino Acid Seauence Analvsis N-terminal amino acid sequence of hIGFBP3 was determined by the application of 530 pmol of recombinant hIGFBP-3 to an Applied Biosytems (ABI) gas-phase sequencer, model 477A. Purification of hIGFBP-3 Recombinant hIGFBP3 was purified from serum-free culture media of CHO cells transfected with an expression vector containing a cDNA for human IGFBP3 (8). Purification of hIGFBP3 from 30 L of CHO conditioned serum-free media was accomplished using ammonium sulphate precipitation and chromatography on heparinagarose, RP-HPLC, and phenyl-Superose. The following buffers were used: Buffer A: 20 mM potassium phosphate, pH 7.0; Buffer B: A + 0.3 M NaCl; Buffer C: A + 0.8 M NaCl; Buffer D: 0.1% TFA, Buffer E: 0.1% TFA in acetonitrile. Protease inhibitors (leupeptin, 1 mM and pepstatin, 1 mM) were added to buffers A, B, C. Steo I. Ammonium Sulfate Precipitation: CHO conditioned medium (30L) was collected and protease inhibitors were added (leupeptin, 1mM; pepstatin, 1mM; phenylmethanesulfonyl fluoride, 100 mM). Culture medium was centrifuged at 15,000 x g for 20 minutes. Supematant was saved and solid ammonium sulfate was added to a final concentration of 60% saturation. The mixture was stirred overnight at 4°C and the precipitate collected by centrifugation at 15,000 x g for 50 min. 2 To be published elsewhere. 626

Vol.

178,

No.

2, 1991

BIOCHEMICAL

AND

BIOPHYSICAL

RESEARCH

COMMUNICATIONS

S ep II. Heparin-Agarose ChromatozraphK Ammonium sulfate precipitate was rtsuspended in 200 ml of 6 M urea containing 20 mM potassium phosphate pH 7 0 0 14% (w/v) protamine sulfate, and protease inhibitors (leupeptin, 1 mM; and pepstat& !l mM), shaken for 15 min at room temperature, and centrifuged at 15,000 x g for 20 min. The supematant was dialyzed against buffer A overnight, centrifuged at 5000 x g for 5 min and loaded onto a heparin-agarose column (5.0 x 14 cm). The column was washed in succession with buffer A and B until the absorbance (280 run) reached baseline and then eluted with buffer C. Sten III. Reverse-Phase HPLC: Proteins eluted from the heparin-agarose column with buffer C (250 ml) were acidified to pH 2.5 with 5% TFA The solution was filtered through a 0.2 urn Millipore low protein binding filter prior to loading onto the HPLC. The Vydac C-4 semiprep column was equilibrated in buffer D and eluted with a linear gradient of O40% buffer E in 40 min (3 ml/mm). The fractions containing hIGFBP-3 were pooled and lyophilized. Steo IV. Phenvl-Superose. FPLC: The lyophilized sample was reconstituted in buffer C and loaded onto a phenyl-Superose HR lO/lO column equilibrated in same buffer. The column was eluted with a linear gradient of O-100% buffer A in 60 ml (1 ml/mm). Tunicamvcin Treatment of Transfected CHO Cells: Stably transfected CHO cells expressing recombinant hIGFBP-3 were grown as previously reported (8). Conditioned medium was assayed (20 ul) for binding activity by ligand-binding. Tunicamycin was added as described by Wieland et al. (14). N-Glycanase Treatment of Purified hIGFBP-3: hIGFBP-3 (50 ug) was desalted by RP-HPLC, dried in a vacuum centrifuge, and resuspended in 10 ul of 0.5% SDS or 0.5% SDS containing 0.1 M 2-mercaptoethanol. Deglycosylation of recombinant hIGFBP-3 was then performed according to the protocol provided by the manufacturer (Genzyme). Samples were run on 10% SDS PAGE under reducing conditions. Determination of the Soecific Bindinp Activitv of hIGFBP3: Purified recombinant hIGFBP-3 was assayed to determine its specific IGF-I binding activity by the method of Lee et al. (15). Reduction and Carboxvmethvlation: Two samples of purified recombinant hIGFBP-3 (one reduced, the other not reduced), were carboxymethylated according to the method of Allen (16).

Purification

of hIGFBP-3:

A typical purification

described in figures l-3 and summarized monitored

from 30 L of CHO conditioned media is in Table I. The purification procedure was

with the ligand binding and western blot assays. Purified hIGFBP3

at - 70°C. A 15% SDS gel was run to analyze the steps in the purification

was stored (Fig. 4). The

Coomassie blue stained gel shows that the phenyl-Superose column pool (Fig.4,lane 6) contained essentially one doublet with a molecular mass of 45/43 kDa. An analytical RPHPLC trace (4.6 x 250 mmj at 215 run of the same protein preparation (Fig. 5) shows one single symmetrical peak. The N-terminal sequence determined for recombinant hIGFBP627

Vol.

178, No. 2, 1991

BIOCHEMICAL

AND BIOPHYSICAL RESEARCH COMMUNICATIONS

1.0

a75. I w ; I f y: 4

0.50

0.25

0

1

M

I

0.3 e i 1 g 0.2 3 % 0.1

02

-

0.0 0

4

8

12

16

20

Retention

24

28

32

36

Time (min)

FjgJ, Heparin-agarox chromatography. Samplewasloadedonto a 5x14 cm columnin 20 mM potassiumphosphate,pH 7.0,and eluted stepwisewith 0.3 M NaCl and 0.8 M NaCl. Shadedregion containshIGFBP-3. E&L Reverse-phaseHPLC of heparin-agarosepurified sample. Heparin-agarosepool appliedto a Vydac C-4 RP-HPLC columnequilibratedwith 0.1% TFA. Proteinswere eluted with a linear gradient of 0.1% TFA/acetonitrile (dotted line). Shadedregion containshIGFBP-3.

WZIS

a2

. 0.5

0.6

E : iz =

0.1

t I a s

a2

0.0

o.0 1

5

10

Xl

Fraction Number

u Phenyl-!hrperosechromatographyof HPIX purified sample.ReconstitutedHPLC pool wasappliedto a HR lO/lO phenyl-Superose columnequilibratedin 20 mM potassium phosphate,pH 7.0 and 0.8 M NaCl. Proteinswere eluted with a linear gradient of 20 mM potassiumphosphate,pH 7.0 (dotted line). Shadedregion containshIGFBP3. 628

40'

Vol.

178,

No.

2, 1991

BIOCHEMICAL

AND

BIOPHYSICAL

TABLE I Summary of Purification of Recombinant hIGFBP3

RESEARCH

COMMUNICATIONS

Expressed in CHO Cells

Total Protein (4

Total hIGFBP-3 (md

30,000

3,000

27

100

60% Ammonium Sulfate Precipitation

200

2,700

25

92

Heparin-agarose Chromatography

250

300

ND

ND

15

4.5

ND

ND

6

2.4

Purification Step

Volume (n.4

Media

RP-HPLC Phenyl-Superose Chromatography

hIGFBP-3 Recovery (%)

2.4

9

‘ND = Not determined 3 was GASSGG, plasma-derived

which agrees with the published N-terminal IGFBP3

Characterization doublet (45/43 kDa)

amino acid sequence of natural

(7,s).

of recombinant

hIGFBP-3:

Recombinant

hIGFBP-3

on a reduced (Fig.6, lane 1) or non-reduced

migrates

as a

(Fig. 4, lane 6) SDS-

M,x10-3 66I 0.4:

4326-

H ij 0.2-

16l4n

VJ

1

2

3

4

5

6

f3 w

oj/' 0

,/’

I! I\ 10

20 Rctentbn

30 40 Tim? (ndn)

60

Eig& 15% SDS gel of steps in hIGFBP-3 purification Lane 1, Molecular weight markers; Lane 2, CHO conditioned medium, 100 pg; Lane 3, resuspended ammonium sulfate pellet, 100 pg; Lane 4, the heparin-agarose column pool, 100 pg; Lane 5, the RPHPLC pool, 6Opg; Lane 6, the phenyl-Superose column pool, 15 pg. Aliquots from each step were applied to a 15% SDS-polyacrylamide gel and run under nonreducing conditions. The gel was stained with Coomassie Blue. J&5,. Reverse-phase HPLC analysis of purified recombinant MGFBP-3. 20pg of phenylSuperose purified material was applied to a Vydac C-4 analytical RP-HPLC column equilibrated in 0.1% TPA/water. Protein was eluted with a linear gradient of 0.1% TPA/acetonitrile (dotted line). (1% acetonitrile/min; flow rate 1 ml/mm). 629

Vol.

178,

No.

BIOCHEMICAL

2, 1991

AND

2

1

BIOPHYSICAL

3

RESEARCH

COMMUNICATIONS

4

Fig. 6. SDS gel of N-Glycanase treated hIGFBP-3. Lane 1, purified recombinant hIGFBP-3,5 pg; Lane 2, purified recombinant hIGFBP-3 treated with N-Glycanase, 5 pg; Lane 3, purified recombinant hIGFBP3 treated with N-Glycanase in the presence of 2mercaptoethanol, 5 pg; Lane 4, E. coli-derived hIGFBP-3, 5 pg. Samples were applied to a 10 % SDS-polyacrylamide gel and run under reducing conditions. The gel was stained with Coomassie Blue. PAGE.

This protein doublet has also been observed in plasma and Cohn fraction IV (6).

Stably transfected CHO cells expressing recombinant tunicamycin, IGF-I

hIGFBP-3,

grown in the presence of

secrete a single 29 kDa protein which maintains its ability to bind radiolabeled

as determined

that the N-linked

by the ligand-binding

carbohydrates

Purified recombinant which cleaves N-linked

assay (data not shown).

are not essential for IGF-I binding

hIGFBP3 carbohydrate

These results suggest

activity.

(Fig. 6. lane 1) was deglycosylated with N-Glycanase at the protein backbone.

treated with N-Glycanase under nonreducing

When the protein was

conditions one major band (37 kDa) was

observed on SDS PAGE (Fig. 6, lane 2). When the protein was treated with N-Glycanase under reducing conditions one major band (30 kDa) was observed on SDS PAGE (Fig. 6, lane 3). Reduction of the protein is therefore required to cleave all the N-linked carbohydrates from the protein backbone with N-Glycanase. The band obtained with NGlycanase under reducing conditions is identical in size to hIGFBP-3 produced from E. coli (Fig. 6, lanes 3,4). It appears that recombinant hIGFBP-3 contains no O-linked sugars since the E. &i-derived hIGFBP-3 migrates identical to the deglycosylated CHO expressed hIGFBP-3 protein on SDS-PAGE under reducing conditions (Fig. 6, lanes 3,4). Purified recombinant hIGFBP3 was assayed to measure the molar binding ratio of hIGFBP3 to IGF-I (15). 18 pmols of IGF-I were precipitated with 20 pmols of hIGFBP3, corresponding to a binding stoichiometry of 1:l. From this data it appears that purified recombinant hIGFBP-3 is fully active and will form a 1:l molar complex with IGF-I in PBS. Recombinant hIGFBP-3 contains 18 cysteines (7,s). Upon reduction, hIGFBP-3 does not bind IGF-I. This lead us to investigate the number of disulfide bonds in hIGFBP3. Two samples of denatured hIGFBP-3, one reduced and one not reduced, were labeled with 630

Vol.

178,

No.

2, 1991

BIOCHEMICAL

AND

BIOPHYSICAL

RESEARCH

COMMUNICATIONS

iodo[2-‘HIacetic acid. The nonreduced sample contained no labeled hIGFBP-3. The reduced sample contained labeled hIGFBP3. 10 pmol of hIGFBP3 were labeled with 184 pmol

of iodoacetic

acid indicating

all 18 cysteines were carboxymethylated

after

reduction.

These results indicate that none of the cysteines in non-reduced hIGFBP-3

are

free to bind iodoacetic acid and might be involved in disulfide bonds. DISCUSSION In this report we present the first purification

and characterization

of recombinant

hIGFBP-3 expressed in CHO cells. The expressed and purified recombinant migrates as a 45/43 kDa doublet on a nonreducing SDS polyacrylamide gel. The recombinant agarose, RP-HPLC,

hIGFBP-3

protein was purified using 3 chromatography

and phenyl-Superose.

hands, the purified recombinant Plasma hIGFBP-3

hIGFBP3

was reported

Sepharose (6). Recombinant

Final yields average approximately

protein

steps: heparin9%. In our

protein is stable when stored at -20” or -70°C.

to be glycosylated since it binds to Concanavalin-A-

hIGFBP3

from CHO cells was also observed to bind to

Concanavalin-A-Sepharose (data not shown). Based on data presented in Fig. 6, it is postulated that all 3 potential N-glycosylation sites (predicted from the cDNA) are occupied in the 45 kDa recombinant hIGFBP-3 protein (upper band of the doublet) and 2 sites are glycosylated in the 43 kDa recombinant hIGFBP-3 protein (lower band of the doublet). Deglycosylation

with N-Glycanase

under non-reducing

conditions

carbohydrate groups from the 45 kDa protein and 1 N-linked

removes 2 N-linked

carbohydrate group from the

43 kDa protein to leave a 37 kDa protein which contains 1 N-linked

carbohydrate group.

Only under reducing conditions were we able to completely remove all carbohydrate groups from hIGFBP3

with N-glycanase and produce a 30 kDa hIGFBP3

identical to the E. coli-derived

hIGFBP-3

protein that migrates

on SDS PAGE under reducing conditions.

There have been other reports (17-19) on the enzymatic deglycosylation of plasma IGFBPs. These experiments were all done with serum or partially purified fractions and the results were visualized by the ligand binding assay. Since IGFBP3

will not bind IGF under

reducing conditions, this excluded the use of reducing agents in the digestions. The lack of reducing agents and a non-glycosylated standard (E. coli-derived IGFBP3) makes it difficult to determine if the IGFBPs in serum were completely deglycosylated. Yang et al. (17) reported the generation of a 34 kDa IGFBP3 from whole and size fractionated adult rat serum after digestion with N-glycanase under nonreducing conditions. Zapf et d (18) deglycosylated rat IGF binding proteins with N-Glycanase under non-reducing conditions and obtained a 37 kDa protein from a triplet (42-49 kDa). Liu et al. (19) reported the generation of a single 31 kDa protein from human serum after digestion of a 41.5/38 kDa IGFBP3 doublet with endoglycosidase-F. Endoglycosidase-F, however, leaves a N-AcetylD-glucosamine

linked to the asparagine. N-Glycanase, on the other hand, cleaves between 631

Vol.

178,

No.

BIOCHEMICAL

2, 1991

the N-Ace@-D-glucosamine

AND

BIOPHYSICAL

RESEARCH

COMMUNICATIONS

and the asparagine. It appears, therefore, that endoglycosidase-

F is able to digest all three N-linked

carbohydrate chains but that N-Glycanase is inhibited

from cleaving one carbohydrate moiety unless the protein is reduced. The reduction of the disulfide bonds appears to allow the N-glycanase access to the N-Acetyl-D-glucosamine bond. The tunicamycin

experiment

formation of the hIGFBP-3/IGF-I serum binding proteins (17,18). IGFBP3

are available,

revealed that carbohydrates are not essential for the complex. Similar findings have been observed with rat Now that both the CHO and E. coli-derived forms of

it will be possible to investigate

the role of the IGFBP3

carbohydrates. In conclusion, homogeneity

recombinant

hIGFBP-3,

in 4 steps. Recombinant

approximately

expressed in CHO cells, has been purified to

hIGFBP-3

is a doublet with a molecular

mass of

45/43 kDa. Only one band (30 kDa) is observed after deglycosylation

with

N-glycanase under reducing and denaturing conditions. hIGFBP3 contains 18 cysteines all of which my be involved in disulfide bonding. The ability to induce and purify quantities of recombinant

hIGFBP-3

from a mammalian

of structure and function of this important

cell source should facilitate further studies

modulator

of IGF action.

ACKNOWLEDGh4ENT!S We would like to thank Jessie Taylor for supplying CHO conditioned media and Armand Carriveau for technical assistance. We also thank Nancy Gerber for preparation of hIGFBP-3 specific antibodies, Rino Lee for the E.coli-derived hIGFBP3, and May Lee for assistance in developing the hIGFBP-3 RIA.

REFERENCES 1. Holly, J.M.P., and Wass, J.A.H. (1989) J. EndocrinoL X22, 611-618 2. Humbel,

R.E. (1990) Eur. .T. Biochem. 190, 445-462

3. Ballard, J., Baxter, R., Binoux, M., Clemmons, D., Drop, S., Hall, K., Hintz, R., Rechler, M., Ruthanen, E., and Schwander, J. (1989) Acru Endocrinol. 121, 751-752 4. Baxter, R.C., and Martin, J.L. (1989) Progress in Growth Factor Research 1, 49-68 5. Baxter, R.C., and Martin, J.L. (1989) Proc. Nad Acud Sci. USA 86, 6898-6902 6. Martin, J.L., and Baxter, R.C. (1986) J. BioL Chem. 261, 8754-8760 7. Wood, W.I., Cachianes, G., Henzel, W.J., Winslow, G.A., Spencer, S.A., Hellmiss, Martin, J.L., and Baxter, R.C. (1988) MoL EndocrzTnoL 2, 1176-1185

R.,

8. Spratt, S.K., Tatsuno, G.P., Yamanaka, M.K., Ark, B.C., Detmer, J., Mascarenhas, D., Flynn, J., Talkington-Verser, C., Spencer, E.M. (1990) Growth Factors 3, 63-72 9. Laemmli,

U.K. (1970) Nature (London) 227, 680-685 632

Vol.

178,

No.

2, 1991

10. Hossenlopp, And

BIOCHEMICAL

AND

P., Seurin, D., Segovia-Q&son,

BIOPHYSICAL

B., Hardouin,

RESEARCH

COMMUNICATIONS

S., and Binoux, M. (1986)

Biochem. 154, 138-143

11. Read, S.M., and Horthcote, 12. Bradford, M.M. (1976) And

D.H. (1981) Anal. Biochem. 116, 53-64 Biochem. 72, 248-254

13. Blum, W.F., Ranke, M.B., Kietzmann, K., Gauggel, E., Zeisel, H.J., and Bierich, J.R. (1990) J. Clin. EndocrinoL Metab. 70, 1292-1298 14. Wieland, F.T., Gleason, M.L., Serafini, T.A., and Rothman, 300

J.E. (1987) Cell 50, 289-

15. Lee, R.Y., Maack, CA., Tressel, T., and Sommer, A. (1991) Abstract, 2nd International IGF Symposium (E. Martin Spencer, Ed), p. 242 16. Allen, G. (1981) Laboratory Techniques in Biochemistry and Molecular Biology (T.S. Workand R.H. Burdon, Ed), Vol9, pp. 30-31. Elsevier/North-Holland Biomedical Press, Amsterdam, The Netherlands 17. Yang, Y.W.H., Wang, J.F., Orlowski, C.C., Nissley, S.P., and Rechler, M.M. Endocrinology 125, 1540-1555

(1989)

18. Zapf, J., Hauri, C., Waldvogel, M., Futo, E., Hasler, H., Binz, K., Guleer, H.P., S&mid, C., and Froesch, E.R. (1989) Proc. N&Z. Acad Sci. USA 86, 3813-3817 19. Liu,F., Powell, D.R., and Hintz, R.L. (1990) J. C&z. Endocrinol. Metub. 70, 620-628

633