PROTEIN
EXPRESSION
AND
PURIFICATION
2, 296-303 (1991)
Expression of Recombinant Human Apolipoprotein A-l in Chinese Hamster Ovary Cells and Escherichia co/i Louise Brissette,* Nicole Cahuzac-Bet,* Yves L. Marcel,* and Eric Rassartt
Marc Desforges,? Jean-Louis
Bet,*
*Laboratoire du Mgtabolisme des Lipoprote’ines, Institut de Recherches Cliniques de Mont&al, Qukbec, Canada H 2 W TDhpartement de Sciences Biologiques, Universite’ du Qubbec ci Mont&al, Qu&ec, Canada H3C 3P8 Received July 19, 1991, and in revised form September
9, 1991
Human recombinant apolipoprotein (apo) A-I was produced by Chinese hamster ovary (CHO) cells and Escherichia coli with expression vectors containing cDNAs encoding preproapoA-I or apoA-I, respectively. The apoA-I from CHO cells was purified from the culture medium by ammonium sulfate precipitation, phenyl-Sepharose chromatography, and aflinity purification on anti-apoA-I immunoabsorber. Human apoA-I was produced in E. coli as a fusion protein with glutathione S-transferase. A four amino acid linker, which separated the two proteins, was specifically recognized and cut by Factor Xa. The purification was accomplished by chromatography of E. coZi extracts on glutathioneSepharose and digestion with Factor Xa. The highest production level was found to be 0.6 rg/ml of culture medium per 48 h for a clone of stable transformant of CHO cells, whereas E. coli could produce as much as 20 aglml of bacterial culture. These apoA-I forms were compared in terms of molecular weight, isoelectric point, and expression of several epitopes. Recombinant apoA-I obtained from CHO cells appears intact and its isoelectric point is compatible with that of the mature form and the proform of apoA-I, whereas a part of the apoA-I produced by E. coli does not contain the COOHterminus. Also, two of six epitopes are expressed to a greater extent in apoA-I obtained from E. coli o 1991 ACTthan in apoA-I obtained from human plasma. demic
Press,
lR7;
triacylglycerols. Apolipoprotein A-I (apoA-I) is a protein of 28 kDa that is a m a jor constituent of the plasma HDL fraction (1). This apolipoprotein is synthesized by hepatic and intestinal cells as a proform named proapoA-I (2). Six a m ino acids, the propeptide, are removed in the circulation by proteolytic cleavage for complete maturation of apoA-I. Both plasma and lymph have been shown to contain a metal-dependent enzyme that processes proapoA-I into mature apoA-I (3). The converting enzyme has also been shown to be synthesized by the hepatoma cell line HepG2 (4). Our laboratory is interested in defining the conformation that apoA-I takes when associated with an HDL particle. Our goal therefore is to eliminate different portions of the a m ino acid sequence to establish their role in the association with lipid complexes. This involves mutagenesis of the apoA-I cDNA and the expression and purification of the mutated apoA-I. Prior to mutagenesis, we conducted a comparison of two different systems of expression that were compatible with convenient subcloning of the mutagenized apoA-I cDNA. In this paper, we report the expression and purification of recombinant human apoA-I from Chinese hamster ovary (CHO) cells and Escherichia cob. W e also compare these recombinant apoA-I with purified human plasma apoA-I.
Inc.
MATERIALS
AND
METHODS
Materials
High-density lipoproteins (HDL)’ are composed of apolipoproteins (ape), phospholipids, cholesterol, and i Abbreviations used: apo, apolipoprotein; SDS, sodium dodecyl sulfate; CHO, Chinese hamster ovary; IEF, isoelectric focusing; PAGE, polyacrylamide gel electrophoresis; HDL, high-density lipoprotein; MEM, minimal essential medium; mAb, monoclonal antibody; PBS, phosphate-buffered saline; RIA, radioimmunoassay; FBS, fetal bovine serum; IPTG, isopropyl-&D-thiogalactopyranoside. 296
Plasmid pGEX-BT, m o lecular biology enzymes, glutathione-Sepharose, phenyl-Sepharose, cyanogen brom ide-activated Sepharose, and IEF ampholytes were purchased from Pharmacia-LKB (Montreal, Quebec). Bluescript plasmids were purchased from Stratagene (La Jolla, CA). Fetal bovine serum (FBS) was obtained from G IBCO (Burlington, Ontario). Human Factor X, Factor X-activating enzyme from venom, culture m e d ia, hygromycin sulfate, LPSR-1 m e d ium, isopropyl-&D1046-5926/91 $3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
EXPRESSION
OF RECOMBINANT
thiogalactopyranoside (IPTG), and electrophoresis quality agarose were obtained from Sigma (St. Louis, MO) while acrylamide, bisacrylamide, and urea came from BDH (Montreal, Quebec). Purification of Recombinant Human apoA-I Expressed by CHO Cells CHO-D426 (5) cells were maintained in minimal essential medium (MEM) supplemented with 100 U/ml penicillin and 10% inactivated FBS at 37°C in a 5% CO2 atmosphere. Cells were co-transfected at 50% confluence by a calcium phosphate precipitate (6) of PY3hygro (7) and PSV2-preproapoA-I (see Results and Discussion) plasmids at a DNA ratio of 1:lO. Selection was conducted by the addition of 200 rglml of hygromycin sulfate for approximately 20 days. This pool of stable transformant CHO cells was analyzed for its ability to express and secrete human apoA-I and was used to isolate high producing clones. Clones were obtained by seeding at a density of 0.5 cell per well (96 multiwell plates) in DME-F12 containing 10% FBS. To produce apoA-I, 30 million cells were seeded in 100 ml of culture medium (DME-F12 containing 5% FBS, 100 U/ml penicillin G, and 100 pg/ml streptomycin sulfate) in a 850-cm2 roller bottle. After 5 days of culture, the medium was harvested every other day and replaced by fresh medium. Purification of apoA-I from the conditioned medium was conducted as follows. First, the medium was subjected to precipitation by addition of a saturated solution of ammonium sulfate to achieve 60% saturation. This mixture was centrifuged for 45 min at 4500 g and the precipitate was resuspended in a minimal volume of buffer A (50 mM Tris, pH 6.6, containing 0.02% NaN,). The sample was loaded onto a phenyl-Sepharose column (10 X 1.5 cm) preequilibrated with buffer A and the column was washed with the same buffer until the absorbance at 280 nm returned to zero. The column was eluted first with buffer A containing 30% propylene glyco1 until the absorbance returned to baseline and then with buffer A containing 70% propylene glycol. The fractions absorbing at 280 nm were pooled and subjected to dialysis against buffer B (15 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.02% NaN,) and loaded on a monoclonal antibody (mAb) 4Hl affinity column. This column was prepared by covalent association of antiapoA-I mAb 4Hl (8) to cyanogen bromide-activated Sepharose as described by the manufacturer. The column was washed with buffer B and apoA-I was eluted with acetic acid (0.1 M). Each fraction was rapidly neutralized by the addition of an adequate volume of a 2 M solution of Tris base. Purification of Recombinant Human apoA-I Expressed by E. coli Human apoA-I was expressed in E. coli as a fusion protein with glutathione S-transferase, essentially as
HUMAN
apoA-I
297
described by Smith and Johnson (9). Briefly, E. coli (Dh5) containing the plasmid pGEX-2T-AI (see Results) were cultured in Terrific Broth (10). At an OD, of 0.5 to 0.75, the expression was initiated with 0.1 mM of IPTG and 3 h later the cells were pelleted, resuspended in phosphate-buffered saline (PBS) containing 1% Triton X-100, and lysed by sonication. The cell lysate was centrifuged at 10,OOOgfor 10 min and the clear supernatant was applied to a 3 X l-cm column of glutathione-Sepharose preequilibrated with PBS containing 1% Triton X-100. The column was washed with 5 bed volumes of PBS and the fusion protein was eluted with 50 mM Tris-HCl, pH 8.0, containing 5 mM reduced glutathione. A pool of fractions absorbing at OD2a, was dialyzed against 10 mM Tris-HCl, pH 8, containing 150 mM NaCl. The fusion protein was digested with activated Factor X as described elsewhere (9) with the exception that the reaction was conducted for 16 h at 20°C. The digestion mixture was then loaded on a glutathione-Sepharose column and apoA-I was recovered in the unretained fraction. Human apoA-I Detection and Characterization Expression of specific epitopes of human apoA-I was accomplished by solid-phase radioimmunoassays (RIA) in the presence of Tween 20 as described by us elsewhere (11). Briefly, Immunolon II Removawells were coated with 0.2 pg of HDL protein in 100 ~1 of 15 mM Na,CO,, pH 9.6, containing 0.02% NaN,, washed, and saturated with 300 ~1 of 0.5% gelatin in PBS, pH 7.2, containing 0.02% NaN,. The monoclonal antibody, at the predetermined optimal dilution, was mixed with serial dilutions of the competing antigen in reaction buffer (0.5% gelatin, 0.05% Tween, 0.02% NaN, in PBS, pH 7.2), added to the coated and saturated wells, and incubated for 1 h at 37°C. The wells were then emptied and washed three times with 0.05% Tween, 0.02% NaN, in PBS, pH 7.2. Finally each well received 100 ~1 of ‘251-labeled rabbit anti-mouse IgG diluted in reaction buffer and was incubated for 1 h at 37°C. After three washes with wash buffer, the wells were counted and the results expressed as BIB,, where B and B, represent the radioactivity bound in the presence and absence, respectively, of competing antigen. ApoA-I was analyzed on conventional 12% SDSpolyacrylamide gel electrophoresis (PAGE) (12) and on vertical slab gel isoelectrofocusing (IEF) (13) using pH 4 to 6 ampholytes. Western blotting and immunodetection were conducted as fully described elsewhere (8,14). Other Methods Protein was determined by the method of Lowry et al. (15). Isolation and purification of human plasma apoA-I and production and purification of mAbs were as described previously (16), with the exception that 3 M
BRISSETTE
298
guanidine-HCl was used instead of 5 M urea to purify plasma apoA-I. Standard techniques of molecular biology were used to construct vectors, to produce and purify plasmids, and to sequence DNA.
ET AL. 600 a 2 a T F 3
RESULTS
AND
Construction
400
300
DISCUSSION
of Plasmids
To create a eucaryotic plasmid capable of expressing human apoA-I, a full apoA-I cDNA was obtained from apoA-I cDNA clones 101 and 121 that were in the PstI site of pBR322 (17). Clone 101 contains sequences starting at the codon specifying amino acid -4 of the propeptide up to a portion of the poly(A) tail of the mRNA, whereas clone 121 contains 96 bp of the 5’noncoding region up to the codon specifying the 224th amino acid. The inserts were excised with Pat1 and further digested with SauI. The most 3’fragment of cDNA clone 101 and the most 5’fragment of cDNA clone 121 were introduced into the PstI site of the plasmid Bluescript SK+ (18). The reconstituted full-length cDNA was excised with HindIII-BumHI enzymes and finally transferred to the PSV2-GPT vector (19) from which the GPT gene had been excised with HindIII-BumHI. This vector containing the entire coding sequence of preproapoA-I was identified as PSVB-preproapoA-I and was used to express human apoA-I in CHO cells. To prepare an expression vector for E. coli, the pGEX-2T vector developed by Smith and Johnson (9) was used. This vector is characterized by a tat promotor (20) followed by the complete sequence of glutathione S-transferase (Sj26) (21) in which a polylinker (BumHI, SmaI, and EcoRI sites) replaces the normal termination site of Sj26. This polylinker is preceded by the sequence encoding the amino acids recognized and cut by thrombin (Leu-Val-Pro-Arg-Gly-Ser). By cloning in this vector, it is possible to express, in E. coli, a desired cDNA as a fusion protein with glutathione S-transferase. However, if the coding sequence does not start with a BamHI site, the expressed protein will contain supplementary amino acids at its NH,-terminus. As this would be the case for human apoA-I, the following cloning was conducted. The cDNA clone 101 insert was excised by P&I, purified, and partially digested by the enzyme Suu3AI. Two oligonucleotides (oligo 1, 5’GATCCATCGAAGGTCGTGACGAGCCACCGCAGAGTCCATGG-3’; oligo 2,3’GTAGCTTCCAGCACTGCTCGGTGGCGTCTCAGGTACCCTAGB) were synthesized to obtain a double-stranded DNA fragment containing, from 5’to 3’, a BamHI adaptor sequence, 16 bp encoding the amino acid sequence recognized and cut by Factor Xa, 24 bp reconstituting the m issing sequence of the early apoA-I cDNA lost with Suu3AI digestion, and finally a Suu3AI adaptor sequence. This fragment was ligated to the Suu3AI-PstI fragment of the apoA-I cDNA together with a BumHI-PstI-cleaved Bluescript
8 2
200
2 P 2
too
3
i
0 2
4
6
8
10
12
Days in culture FIG. 1_. Effect of different culture media on the production of recombinant human apoA-I by CHO cells at different days of culture. At Day 0, apoA-I-CH04 cells (1.5 X 106) were seeded in lo-cm’ dishes containing 4 ml MEM supplemented with 5% FBS. At Day 2, the cells were approximately 90% confluent and the medium was replaced by MEM containing 5% FBS (I), 2% LPSR-I (f& ITS alone (O), or ITS with 1% FBS (8). Every other day, the production of recombinant human apoA-I was determined by solid-phase RIA and the medium was replaced.
plasmid (KS+). Positive clones were confirmed by sequencing and the insert was excised with BarnHI and EcoRI enzymes, purified, and introduced in a pGEX-2T plasmid digested with the same enzymes. The expression of this clone, identified as pGEX-BT-AI, together with a factor Xa digestion results in a fully mature apoA-I. Production and Purification of Recombinant upoA-I Produced in CHO Cells
Human
The presence of apoA-I in the culture medium of a pool of stable transformants of confluent CHO cells was estimated by solid-phase RIA at 0.1 t 0.05 pg/ml of culture medium (mean f SE, n = 4) for 2 days. High producer clones were isolated by lim iting dilution. After the screening of 480 clones, 3 producing 0.2 pg/ml and 1 producing 0.5 pg/ml of apoA-I were found. This cellular clone was designated apoA-I-CH04 and was used in the studies described below. Production of apoA-I was followed with different culture media to choose the best medium for production and purification. Figure 1 shows the concentration of human apoA-I in the medium at times ranging from 2 to 12 days of culture in DME-F12 medium containing either 5% FBS or 2% of a serum substitute (LPSR-I), or ITS (3 mg/liter bovine insulin, 3 mg/liter human transferrin, and 5 pg/liter sodium selenite) (22,23) alone or with 1% FBS. As shown, the production of apoA-I increases with longer periods of culture, with the exception of when ITS-supplemented medium is used. This medium induces a decline of production at Day 4 and
EXPRESSION TABLE Purification
of Recombinant Volume
Fraction Conditioned
medium
b&d
317
apoA-I
from
apA-
apOA-I yield (W)
bd
2250
CHO
apoA-I
299
Cells
Fold purification -
29 19
7OaO
162 107
100
Precipitation Phenyl-Sepharose
59
2
250
42
23
33
Affinity chromatography
1
96
6
415
10.4
Note. Recombinant human apoA-I was produced by apoA-I-CH04 cells cultured in MEM supplemented with 5% FBS and purified. Samples obtained at different steps of the purification were assayed for protein content by the Lowry method (15) and human apoA-I was determined by RIA (11).
from that day the production of human apoA-I is lower than that with the other media. The addition of 1% FBS to the ITS-supplemented medium improved production but it remained consistently lower than that with the medium containing 5% FBS. A similar production was also observed with the medium supplemented with 2% LPSR-1. The use of LPSR-1 was excluded since we found that this medium contains approximately as much bovine apoA-I as 5% FBS-supplemented medium. Thus, the medium containing 5% FBS was chosen for subsequent studies because of the level of production and the lack of an acceptable alternative. Purification of recombinant apoA-I from culture medium consisted of chromatography on phenyl-sepha-
A
HUMAN
1
Human Protein
(ml)
OF RECOMBINANT
B
1
234
1
12
5
FIG. 3.
Isoelectric focusing of recombinant human apoA-I produced by CHO cells and by E. coli. Recombinant human apoA-I was produced by apoA-I-CH04 cells and E. coli (pGEX-PT-AI) and the charges of these different preparations of apoA-I were analyzed by isoelectric focusing on a gel containing ampholytes from pH 4 to 6. At the completion of the electrophoresis, the proteins were transferred to nitrocellulose and apoA-I was detected with anti-human apoA-I mAb A05 (A), mAb 4Hl (B), or mAb 4A12 (C). (A) Lane 1, bovine serum; lanes 2 to 4, different preparations of recombinant human apoA-I produced by CHO cells; lane 5, human plasma apoA-I. (B) Lane 1, human plasma apoA-I; lane 2, recombinant apoA-I produced by E. coli. (C) Lane 1, recombinant apoA-I produced by E. coli. The arrows indicate the position of the mature form of apoA-I.
rose followed by immunoaffinity chromatography on an anti-apoA-I mAb 4Hl-Sepharose. This immunoabsorber was chosen because mAb 4Hl does not crossreact with bovine apoA-I (manuscript in preparation). This was also verified by RIA and by applying bovine serum onto a 4Hl immunoabsorber, which, respectively, resulted in no detection and no retention (data
44.2 27.0 18.3 -
12345
12
FIG. 2. Electrophoretic pattern of recombinant human apoA-I produced by CHO cells at different steps of the purification. Recombinant human apoA-I was produced by apoA-I-CH04 cells cultured in MEM supplemented with 5% FBS and purified. Samples obtained at different steps of the purification were NII on SDS-PAGE (12%) and either stained by Coomassie blue (A) or transferred to nitrocellulose and detected by a mixture of anti-human apoA-I mAbs (4H1, 5F6, 4A12) (B). (A) Lane 1, conditioned medium; lane 2, ammonium sulfate precipitate; lane 3, sample after phenyl-Sepharose chromatography; lane 4, sample after afEnity chromatography on mAb 4Hl-Sepharose; lane 5, human plasma apoA-I. (B) Lane 1, human plasma apoA-I; lane 2, purified recombinant human apoA-I.
123456
FIG. 4.
12
3
4
5
12345
Electrophoretic pattern of E. coli proteins at different steps of the purification. Recombinant human apoA-I was produced by E. coli (pGEX-2T-AI) and purified. The samples obtained at different steps of the purification were run on SDS-PAGE (12%) and either stained by Coomassie blue (A) or transferred to nitrocellulose and probed with anti-human apoA-I mAb 4Hl (B) or mAb 4A12 (C). (A, B, and C) Lane 1, E. coli cell extract; lane 2, cell extract purified by glutathione-Sepharose affinity chromatography; lane 3, purified extract digested by Factor Xa; lane 4, fraction of the digest not retained by glutathione-Sepharose affinity chromatography. (A) Lane 5, twice the amount of protein applied in lane 4; lane 6, purified human plasma apoA-I. (B and C) Lane 5, purified human plasma apoA-I.
300
BRISSETTE
not shown). The results of purification are summarized in Table 1 and the electrophoretic pattern of samples taken at different steps of the purification can be seen in Fig. 2A. The major protein found in the culture medium is a 65kDa protein presumed to be bovine albumin of the FBS (Fig. 2A, lane 1). As purification proceeds, a protein of 28 kDa appears (Fig. 2A, lane 3). This band comigrates with human serum apoA-I and coincides with a protein detectable on a Western blot with mAb anti-human apoA-I (Fig. 2B). Judged by the intensities of the band, the protein of 28 kDa represents more than 10% of the proteins purified by phenyl-Sepharose and more than 90% of the proteins detected by Coomassie blue staining of the SDS-PAGE, at the end of purification (Fig. 2A, lane 4). However, the contribution of apoA-I determined by RIA suggests that human apoA-I contributes to approximately 1 and 11% of the protein content of the preparations obtained after purification on phenyl-Sepharose and on mAb 4Hl-Sepharose, respectively (Table 1). This discrepancy between the two methods may be a consequence of the immunodetection of recombinant apoA-I produced by CHO cells being weaker than that of plasma apoA-I, as shown and discussed in the last part of the manuscript. The apoA-I is secreted as a molecule of proapoA-I which loses its prosegment of 6 amino acids to become a fully mature apoA-I (2). The proapoA-I usually represents less than 5% of plasma human apoA-I and its isoelectric point is 5.85 instead of 5.65 (24). Figure 3A compares the IEF patterns of three different preparations of purified recombinant human apoA-I with apoA-I purified from human plasma HDL. Thus, the preparations of recombinant apoA-I (lanes 2 to 4) are richer in a molecule with an isoelectric point comparable with human plasma proapoA-I (lane 5) than in the presumed mature form (approximately 30%). This experiment was conducted with mAb A05, which cross-reacts with bovine apoA-I (manuscript in preparation), but similar experiments with mAbs that do not recognize bovine apoA-I gave essentially the same results. The mAb A05 was chosen because bovine apoA-I was analyzed in the same experiment. Figure 3A also shows that the isoelectric point of bovine apoA-I (lane 1) is higher than that of the mature and the proform of recombinant human apoA-I. This corroborates the specificity of the 4Hl affinity column used for purification of human apoA-I, since no signal of an apoA-I molecule with this isoelectric point is detectable in the preparation of purified recombinant human apoA-I. Production and Purification of Recombinant Human apoA-I Produced by E. coli Approximately 20 mg of recombinant human apoA-I could be produced per liter of E. coli culture. Figure 4A shows an electrophoretic pattern of the proteins ob-
ET AL. TABLE 2 Purification
of Human Volume
Fraction Cell extract Affinity chromatography
(ml)
Recombinant
from E. coli
apoA-I
Protein
apoA-I
Wml)
Wml)
apoA-I yield (7%)
Fold purification
14
23,000
3ooo
100
-
20
!m
720
34
6.3
Note. Recombinant human apoA-I was produced by E. coli (pGEXST-AI) and purified. Samples obtained at different steps of the purification were assayed for protein content by the Lowry method (15) and human apoA-I was determined by RIA (11).
tained at different steps of the purification and Table 2 shows the extent of purification achieved. The glutathione-Sepharose affinity column retains a protein with an apparent molecular weight of 55 kDa, as expected for a glutathione S-transferase-apoA-I fusion protein (lane 2). Three other bands of molecular weight ranging from 30 to 35 kDa but with intensities lower than that of the 55-kDa protein are also observed. These preparations were assessed by immunoblotting with mAb 4H1, which recognizes an epitope close to the NH,-terminus (8), and mAb 4A12, which is specific for an epitope located in the last third of the apoA-I sequence (8). Figure 4C (lane 2) shows that mAb 4A12 recognizes a 55-kDa protein and a family of proteins of molecular weight ranging from 30 to 35 kDa, as well as some low-molecularweight peptides. Thus, the immunodetection of the purified fusion protein with mAb 4A12 and staining with Coomassie blue give similar patterns. This suggests that the family of proteins of size ranging from 30 to 35 kDa is composed of the first two-thirds of the apoA-I molecule linked to a terminal portion of the glutathione Stransferase molecule large enough to be retained by glutathione-Sepharose. A different pattern is obtained (Fig. 4B, lane 2) when a mAb recognizing the extreme NH,-terminal region of apoA-I is used for detection. Indeed, not only the bands detected with mAb 4A12 but also a family of proteins of 10 kDa and several less intense signals distributed between 55 and 10 kDa were observed. This implies that an important fraction of the fusion proteins contains the NH,-terminus of apoA-I but not the COOH-terminus. Furthermore, the more important contribution of these multiple bands on immunoblots than on Coomasie blue-stained gels suggests that the smaller fusion proteins are more immunoreactive and/or are less chromogenic than the full-size fusion protein. Several attempts were made to eliminate these multiple bands. Expression of apoA-I was determined at different times (0 to 3 h) after the induction with IPTG, different concentrations of IPTG were tested, and conventional protease inhibitors were added. Also, other
EXPRESSION
Bmmax
OF RECOMBINANT
4Hl
Bmmax IPJ
‘*el
Concentration
I
BIBmax
, 2 B/Bmax
3610
All
@g/nG
5F6 1.0 B’Bmax
I
,2 BfBmax
4A12
I
I
Concentration FIG. 5. Competitive protein was competed from E. coli (0).
301
apoA-I
(pg/ml)
2611
¢radon
HUMAN
@g/ml)
solid-phase radioimmunoassays. The interaction of different mAbs (4H1,2Gll, 3G10, All, 5F6, and 4A12) with HDL with purified human plasma apoA-I (A) or with purified recombinant human apoA-I obtained from CHO cells (Cl) or
procedures for protein extraction were tested (8) including one selecting for cytoplasmic and periplasmic proteins (25). E. coli POP-2136 that are protease deficient were also used as a host cell. Under all these conditions, there was no clear improvement (data not shown), suggesting that gene induction and proteolysis secondary to the breakage of the cells are probably not responsible for the smaller products detected. Consequently, this suggests that either transcription or translation is prematurely interrupted. Digestion of the fusion protein with Factor Xa gives a doublet of bands with an apparent molecular weight of 28 kDa on SDS-PAGE (Fig. 4A, lane 3). The smaller band comigrates with human apoA-I (lane 6) while the larger one is likely glutathione S-transferase as judged by its disappearance upon passage on glutathioneSepharose (lane 4). Several less intense bands of molec-
ular weight lower than 28 kDa could also be observed. In accordance with the results obtained with the fusion protein, immunodetection with mAb 4Hl revealed several bands in the digested sample while only bands of 28 kDa and of less than 10 kDa were detectable with mAb 4A12 (Figs. 4B and 4C, respectively, lanes 4). The isoelectric point of this recombinant human apoA-I was analyzed by immunodetection of Western blots of IEF gels. Figure 3C (lane 1) shows that mAb 4A12 detects a protein that comigrates with the mature form of plasma apoA-I. However, mAb 4Hl (Fig. 3B, lane 2) recognizes not only this apoA-I but also a series of molecules with lower isoelectric point. Together the immunodetection of Western blots of SDS-PAGE and IEF suggests that the intact apoA-I molecule produced by E. coli has the same isoelectric point as human plasma apoA-I.
302
BRISSETTE TABLE
3
Immunoreactivity of Human apoA-I Produced by CHO Cells and E. coli in Comparison with Purified Human Plasma apoA-I ED,
recombinant
mAb
CHO cells
4Hl 3G10 2Gll All 5F6 4Al2 N
2.2 3.4 6.7 2.1 7.6 5.5
f + f + f f
0.75” 1.6 2.1 0.05 2.4 0.85 2
apoA-I/ED, apoA-I
plasma
E. coli 0.12 3.7 8.5 0.08 8.7 7.6
f + f + + f
o.007b 1.4 4.3 0.009’ 3.6 0.88 3
Note. Recombinant human apoA-I obtained from CHO cells or E. coli was tested in competitive RIA with the mAbs indicated. ED, is the effective dose at 50% binding measured by RIA in the presence of Tween 20. a Values represent the mean ? SE. b P < 0.05. c P < 0.0001.
Comparison of the Expression of Epitopes between Recombinant Human apoA-I and Purified Human Plasma apoA-I
The characterization of the different recombinant human apoA-I was completed by an analysis of the expression of six epitopes distributed in the primary sequence of the apoA-I molecule. This was accomplished by competitive solid-phase RIA using different mAbs, including mAb 4H1, which recognizes an epitope mapped between amino acids 2 and 8; mAb 2Gl1, which is specific for an epitope determined between amino acids 44 and 112; and mAbs 2G10, All, 5F6, and 4A12, which recognize epitopes located between amino acids 98 and 121,99 and 132,118 and 148, and 173 and 205, respectively (8). Two preparations of human apoA-I obtained from CHO cells and three preparations from E. coli were tested with three preparations of purified human plasma apoA-I. The displacement curves for all rnAbs are shown in Fig. 5 for one preparation of each type of apoA-I and the mean and standard deviation of the ratios of the ED, (effective dose at 50% binding) of all preparations to the ED, of purified human plasma apoA-I appear in Table 3. Figure 5 shows that independent of the mAb used, the slopes of the competition curves are identical among the different preparations of apoA-I. This indicates that the aflinities of the different monoclonal antibodies for purified plasma apoA-I and recombinant human apoA-I are similar and suggests a similar conformation of the epitopes in the various apoA-I forms tested. However, for most RIA (2Gl1, 3Gl0, 5F6, 4A12), a shift in the curves toward higher
ET
AL.
apoA-I concentrations is observed with recombinant human apoA-I produced by either CHO cells or E. coli. Table 3 shows that the ED,, ratios determined with mAbs 2Gl1, 3G10, 5F6, and 4A12 are not significantly different between the recombinant human apoA-I forms. In these assays, a concentration of the two types of recombinant human apoA-I much higher (2.1- to 7.6fold) than that of purified plasma apoA-I is needed to achieve half the maximal binding with the mAbs. Thus, it can be hypothesized that either the contribution of recombinant apoA-I in the purified preparations has been overestimated or a fraction of the purified recombinant apoA-I has been modified in the process of purification. The second hypothesis must be rejected because we verified that the isolation of human plasma apoA-I by the methods used for CHO-derived apoA-I did not alter the immunoreactivity of apoA-I (data not shown). With E. coli-derived preparations, there is a net and significant shift toward apparent higher concentrations in apoA-I epitopes recognized by mAbs 4Hl and All (Fig. 5 and Table 3). This can be explained by the more abundant number of recombinant apoA-I molecules containing the NH,-terminus than the COOH-terminus and also by the higher immunoreactivity of smaller fragments of apoA-I. The reasons the epitopes of mAbs 2Gll and 3GlO mapped between the epitopes of mAbs 4Hl and All did not show a similar shift remain obscure. Thus, a different folding between human apoA-I produced in CHO cells and in E. coli cannot be excluded. CONCLUSION
In conclusion, we have produced and purified recombinant human apoA-I from procaryotic and eucaryotic systems and compared these apoA-I forms in terms of molecular weight, isoelectric point, and expression of several epitopes. Although the CHO stable transformant cells give a low yield of production, the recombinant apoA-I appears intact and its isoelectric point is compatible with that of the mature and the proform of apoA-I normally found in plasma. Despite the high yield of production of the E. coli system, part of the apoA-I produced is incomplete in the COOH-terminus. Furthermore, in contrast to the recombinant human apoA-I produced by CHO cells, two of six epitopes are expressed to a greater extent in apoA-I obtained from E. coli than from human plasma. Thus, the CHO system appears to be the best system to produce human recombinant apoA-I. ACKNOWLEDGMENTS We thank Dr. V. man apoA-I cDNA treal) for the CHO Milne for his useful We also gratefully
Zannis (Boston University) for providing the huand Dr. W. E. C. Bradley (Cancer Institute, Moncells. We are indebted to our colleague Dr. R. W. suggestions in the preparation of the manuscript. acknowledge the help of Ms. C. Lemire in typing
EXPRESSION
OF RECOMBINANT
the manuscript. This work was supported by the Medical Research Council of Canada (PG-2’7). L.B. is a recipient of a scholarship from the Canadian Heart Foundation.
REFERENCES 1. Swaney, J. B., Braithwaite, F., and Eder, H. A. (1977) Characterization of the apolipoproteins of rat plasma lipoproteins. Biochem-
i&y16,271-278. 2. Gordon, J. I., Smith, D. P., Andy, R., Alpers, D. H., Schonfeld,
3.
4.
5.
6.
7.
8.
9.
10. 11.
12.
G., and Strauss, A. W. (1982) The primary translation product of rat intestinal apolipoprotein mRNA is an unusual preproprotein. J. Biol. Chem. 257,971-978. Edelstein, C., Gordon, J. I., Toscas, K., Sims, H. F., Strauss, A. W., and Scanu, A. M. (1983) In vitro conversion of proapoprotein A-I to apoprotein A-I. J. Biol. Chem. 268, 11,430-11,433. Edelstein, C., Kaiser, M., Piras, G., and Scanu, A. M. (1988) Demonstration that the enzyme that converts precursor of apolipoprotein A-I to apolipoprotein A-I is secreted by the hepatocarcinoma cell line HepG2. Arch. Biochem. Biophys. 267,23-30. Belouchi, A., and Bradley, W. E. C. (1991) Analysis of second step mutation of class II and class III CHO APRT gene heterozygote: Chromosomal differences and dilution frequencies. Somatic Cell Mol. Genet., 17,277-286. Wigler, M., Pellicer, A., Silvertein, S., and Axel, R. (1978) Biochemical transfer of single-copy eucaryotic genes using total cellular DNA as donor. Cell 14, 725-731. Blocklinger, K., and Digglemann, H. (1984) Hygromycin B transferase as a stable marker for DNA transfer experiments with higher eucaryotic cells. Mol. Cell Biol. 4, 2929-2931. Marcel, Y. L., Provost, P. R., Koa, H., Raffai, E., Dac, N. V., Fruchart, J.-C., and Rassart, E. (1991) The epitopes of apolipoprotein A-I define distinct structural domains including a mobile middle region. J. Biol. Chem. 266,3644-3653. Smith, D. B., and Johnson, K. S. (1988) Single-step purification of polypeptides expressed in Escherichia coli as fusions with glutsthione S-transferase. Gene 6’7, 31-40. Tartof, K. D., and Hobbs, C. A. (1987) Improved media for growing plasmid and cosmid clones. Bethesda Res. Lab. Focus 9,12-13. Marcel, Y. L., Jewer, D., Vezina, C., Milthorp, P., and Weech, P, K. (1987) Expression of human apolipoprotein A-I epitopes in high density lipoproteins and in serum. J. Lipid Res. 28,768-777. Laemmli, U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227,680-685.
HUMAN
apoA-I
303
electro13. O’Farrell, P. H. (1975) High resolution two-dimensional phoresis of proteins. J. Biol. Chem. 250,4007-4021. 14. Marcel, Y. L., Jewer, D., Leblond, L., Weech, P. K., and Milne, R. W. (1989) Lipid peroxidation changes the expression of specific epitopes of apolipoprotein A-I. J. Biol. C&em. 264, 19,94219,950. 15. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) Protein measurement with the folin phenol reagent. J. Biol. Chem. 193,265-275. 16. Brewer, H. B., Ronan, R., Meng, M., and Bishop, C. (1986) Isolation and characterization of apolipoproteins A-I, A-II, and A-IV, in “Methods in Enzymology” (Segrest, J. P., and Albers, J. J., Eds.), Vol. 128, pp. 223-246, Academic Press, San Diego. S. K., Zannis, V. I., and Breslow, J. L. (1983) Isola17. Karathanasis, tion and characterization of the human apolipoprotein A-I gene. Proc. Natl. Acad. Sci. USA 80,6147-6151. 18. Short, J. M., Fernandez, J. M., Sorge, J. A., and Huse, W. D. (1988) Lambda ZAP: A bacteriophage lambda expression vector with in vivo excision properties. Nucleic Acids Res. 16, 7583-
7600. 19. Mulligan, R. C., and Berg, P. (1981) Selection for animal cells that express the Escherichia coli gene coding for xanthine-guanine phosphoribosyltransferase. Proc. Natl. Acad. Sci. USA 78,
2072-2076. 20. Amann, E., Brosius, J., and Ptashne, M. (1983) Vectors bearing a hybrid trplac promotor useful for regulated expression of cloned genes in Escherichia coli. Gene 25,167-178. 21. Smith, D. B., Davern, K. M., Board, P. G., Tiu, W. U., Garcia, E. G., and Mitchell, G. F. (1986) Mr 26 000 antigen of Schistosoma japonicum recognized by resistant WEHI 12915 mice is a parasite glutathione S-transferase. Proc. Natl. Acad. Sci. USA 83,97039707. 22. Hamilton, W. G., and Ham, R. G. (1977) Clonal growth of Chinese hamster cell lines in protein-free media. In Vitro 13, 537-
547.
23
Gasser, F., Mulsant, P., and Gillois, M. (1987) Long term multiplication of the Chinese hamster ovary (CHO) cell line in a serum-free medium. In Vitro Cell. Deu. Biol. 21, 588-592. 24 Zannis, V. I., Breslow, J. L., and Katz, A. J. (1980) Isoproteins of ’ human apolipoprotein A-I demonstrated in plasma and intestinal organ culture. J. Biol. Chem. 255,8612-8617. 25. Rozenwasser, T. A., Hogquist, K. A., Nothwehr, S. F., BradfordGoldberg, S., Olins, P. O., Chaplin, D. D., and Gordon, J. I. (1990) Compartmentalization of mammalian proteins produced in Escherichia coli. J. Biol. Chem. 265. 13,066-13,073. ’
EXPRESSION
AND
PURIFICATION
2, 296-303 (1991)
Expression of Recombinant Human Apolipoprotein A-l in Chinese Hamster Ovary Cells and Escherichia co/i Louise Brissette,* Nicole Cahuzac-Bet,* Yves L. Marcel,* and Eric Rassartt
Marc Desforges,? Jean-Louis
Bet,*
*Laboratoire du Mgtabolisme des Lipoprote’ines, Institut de Recherches Cliniques de Mont&al, Qukbec, Canada H 2 W TDhpartement de Sciences Biologiques, Universite’ du Qubbec ci Mont&al, Qu&ec, Canada H3C 3P8 Received July 19, 1991, and in revised form September
9, 1991
Human recombinant apolipoprotein (apo) A-I was produced by Chinese hamster ovary (CHO) cells and Escherichia coli with expression vectors containing cDNAs encoding preproapoA-I or apoA-I, respectively. The apoA-I from CHO cells was purified from the culture medium by ammonium sulfate precipitation, phenyl-Sepharose chromatography, and aflinity purification on anti-apoA-I immunoabsorber. Human apoA-I was produced in E. coli as a fusion protein with glutathione S-transferase. A four amino acid linker, which separated the two proteins, was specifically recognized and cut by Factor Xa. The purification was accomplished by chromatography of E. coZi extracts on glutathioneSepharose and digestion with Factor Xa. The highest production level was found to be 0.6 rg/ml of culture medium per 48 h for a clone of stable transformant of CHO cells, whereas E. coli could produce as much as 20 aglml of bacterial culture. These apoA-I forms were compared in terms of molecular weight, isoelectric point, and expression of several epitopes. Recombinant apoA-I obtained from CHO cells appears intact and its isoelectric point is compatible with that of the mature form and the proform of apoA-I, whereas a part of the apoA-I produced by E. coli does not contain the COOHterminus. Also, two of six epitopes are expressed to a greater extent in apoA-I obtained from E. coli o 1991 ACTthan in apoA-I obtained from human plasma. demic
Press,
lR7;
triacylglycerols. Apolipoprotein A-I (apoA-I) is a protein of 28 kDa that is a m a jor constituent of the plasma HDL fraction (1). This apolipoprotein is synthesized by hepatic and intestinal cells as a proform named proapoA-I (2). Six a m ino acids, the propeptide, are removed in the circulation by proteolytic cleavage for complete maturation of apoA-I. Both plasma and lymph have been shown to contain a metal-dependent enzyme that processes proapoA-I into mature apoA-I (3). The converting enzyme has also been shown to be synthesized by the hepatoma cell line HepG2 (4). Our laboratory is interested in defining the conformation that apoA-I takes when associated with an HDL particle. Our goal therefore is to eliminate different portions of the a m ino acid sequence to establish their role in the association with lipid complexes. This involves mutagenesis of the apoA-I cDNA and the expression and purification of the mutated apoA-I. Prior to mutagenesis, we conducted a comparison of two different systems of expression that were compatible with convenient subcloning of the mutagenized apoA-I cDNA. In this paper, we report the expression and purification of recombinant human apoA-I from Chinese hamster ovary (CHO) cells and Escherichia cob. W e also compare these recombinant apoA-I with purified human plasma apoA-I.
Inc.
MATERIALS
AND
METHODS
Materials
High-density lipoproteins (HDL)’ are composed of apolipoproteins (ape), phospholipids, cholesterol, and i Abbreviations used: apo, apolipoprotein; SDS, sodium dodecyl sulfate; CHO, Chinese hamster ovary; IEF, isoelectric focusing; PAGE, polyacrylamide gel electrophoresis; HDL, high-density lipoprotein; MEM, minimal essential medium; mAb, monoclonal antibody; PBS, phosphate-buffered saline; RIA, radioimmunoassay; FBS, fetal bovine serum; IPTG, isopropyl-&D-thiogalactopyranoside. 296
Plasmid pGEX-BT, m o lecular biology enzymes, glutathione-Sepharose, phenyl-Sepharose, cyanogen brom ide-activated Sepharose, and IEF ampholytes were purchased from Pharmacia-LKB (Montreal, Quebec). Bluescript plasmids were purchased from Stratagene (La Jolla, CA). Fetal bovine serum (FBS) was obtained from G IBCO (Burlington, Ontario). Human Factor X, Factor X-activating enzyme from venom, culture m e d ia, hygromycin sulfate, LPSR-1 m e d ium, isopropyl-&D1046-5926/91 $3.00 Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
EXPRESSION
OF RECOMBINANT
thiogalactopyranoside (IPTG), and electrophoresis quality agarose were obtained from Sigma (St. Louis, MO) while acrylamide, bisacrylamide, and urea came from BDH (Montreal, Quebec). Purification of Recombinant Human apoA-I Expressed by CHO Cells CHO-D426 (5) cells were maintained in minimal essential medium (MEM) supplemented with 100 U/ml penicillin and 10% inactivated FBS at 37°C in a 5% CO2 atmosphere. Cells were co-transfected at 50% confluence by a calcium phosphate precipitate (6) of PY3hygro (7) and PSV2-preproapoA-I (see Results and Discussion) plasmids at a DNA ratio of 1:lO. Selection was conducted by the addition of 200 rglml of hygromycin sulfate for approximately 20 days. This pool of stable transformant CHO cells was analyzed for its ability to express and secrete human apoA-I and was used to isolate high producing clones. Clones were obtained by seeding at a density of 0.5 cell per well (96 multiwell plates) in DME-F12 containing 10% FBS. To produce apoA-I, 30 million cells were seeded in 100 ml of culture medium (DME-F12 containing 5% FBS, 100 U/ml penicillin G, and 100 pg/ml streptomycin sulfate) in a 850-cm2 roller bottle. After 5 days of culture, the medium was harvested every other day and replaced by fresh medium. Purification of apoA-I from the conditioned medium was conducted as follows. First, the medium was subjected to precipitation by addition of a saturated solution of ammonium sulfate to achieve 60% saturation. This mixture was centrifuged for 45 min at 4500 g and the precipitate was resuspended in a minimal volume of buffer A (50 mM Tris, pH 6.6, containing 0.02% NaN,). The sample was loaded onto a phenyl-Sepharose column (10 X 1.5 cm) preequilibrated with buffer A and the column was washed with the same buffer until the absorbance at 280 nm returned to zero. The column was eluted first with buffer A containing 30% propylene glyco1 until the absorbance returned to baseline and then with buffer A containing 70% propylene glycol. The fractions absorbing at 280 nm were pooled and subjected to dialysis against buffer B (15 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.02% NaN,) and loaded on a monoclonal antibody (mAb) 4Hl affinity column. This column was prepared by covalent association of antiapoA-I mAb 4Hl (8) to cyanogen bromide-activated Sepharose as described by the manufacturer. The column was washed with buffer B and apoA-I was eluted with acetic acid (0.1 M). Each fraction was rapidly neutralized by the addition of an adequate volume of a 2 M solution of Tris base. Purification of Recombinant Human apoA-I Expressed by E. coli Human apoA-I was expressed in E. coli as a fusion protein with glutathione S-transferase, essentially as
HUMAN
apoA-I
297
described by Smith and Johnson (9). Briefly, E. coli (Dh5) containing the plasmid pGEX-2T-AI (see Results) were cultured in Terrific Broth (10). At an OD, of 0.5 to 0.75, the expression was initiated with 0.1 mM of IPTG and 3 h later the cells were pelleted, resuspended in phosphate-buffered saline (PBS) containing 1% Triton X-100, and lysed by sonication. The cell lysate was centrifuged at 10,OOOgfor 10 min and the clear supernatant was applied to a 3 X l-cm column of glutathione-Sepharose preequilibrated with PBS containing 1% Triton X-100. The column was washed with 5 bed volumes of PBS and the fusion protein was eluted with 50 mM Tris-HCl, pH 8.0, containing 5 mM reduced glutathione. A pool of fractions absorbing at OD2a, was dialyzed against 10 mM Tris-HCl, pH 8, containing 150 mM NaCl. The fusion protein was digested with activated Factor X as described elsewhere (9) with the exception that the reaction was conducted for 16 h at 20°C. The digestion mixture was then loaded on a glutathione-Sepharose column and apoA-I was recovered in the unretained fraction. Human apoA-I Detection and Characterization Expression of specific epitopes of human apoA-I was accomplished by solid-phase radioimmunoassays (RIA) in the presence of Tween 20 as described by us elsewhere (11). Briefly, Immunolon II Removawells were coated with 0.2 pg of HDL protein in 100 ~1 of 15 mM Na,CO,, pH 9.6, containing 0.02% NaN,, washed, and saturated with 300 ~1 of 0.5% gelatin in PBS, pH 7.2, containing 0.02% NaN,. The monoclonal antibody, at the predetermined optimal dilution, was mixed with serial dilutions of the competing antigen in reaction buffer (0.5% gelatin, 0.05% Tween, 0.02% NaN, in PBS, pH 7.2), added to the coated and saturated wells, and incubated for 1 h at 37°C. The wells were then emptied and washed three times with 0.05% Tween, 0.02% NaN, in PBS, pH 7.2. Finally each well received 100 ~1 of ‘251-labeled rabbit anti-mouse IgG diluted in reaction buffer and was incubated for 1 h at 37°C. After three washes with wash buffer, the wells were counted and the results expressed as BIB,, where B and B, represent the radioactivity bound in the presence and absence, respectively, of competing antigen. ApoA-I was analyzed on conventional 12% SDSpolyacrylamide gel electrophoresis (PAGE) (12) and on vertical slab gel isoelectrofocusing (IEF) (13) using pH 4 to 6 ampholytes. Western blotting and immunodetection were conducted as fully described elsewhere (8,14). Other Methods Protein was determined by the method of Lowry et al. (15). Isolation and purification of human plasma apoA-I and production and purification of mAbs were as described previously (16), with the exception that 3 M
BRISSETTE
298
guanidine-HCl was used instead of 5 M urea to purify plasma apoA-I. Standard techniques of molecular biology were used to construct vectors, to produce and purify plasmids, and to sequence DNA.
ET AL. 600 a 2 a T F 3
RESULTS
AND
Construction
400
300
DISCUSSION
of Plasmids
To create a eucaryotic plasmid capable of expressing human apoA-I, a full apoA-I cDNA was obtained from apoA-I cDNA clones 101 and 121 that were in the PstI site of pBR322 (17). Clone 101 contains sequences starting at the codon specifying amino acid -4 of the propeptide up to a portion of the poly(A) tail of the mRNA, whereas clone 121 contains 96 bp of the 5’noncoding region up to the codon specifying the 224th amino acid. The inserts were excised with Pat1 and further digested with SauI. The most 3’fragment of cDNA clone 101 and the most 5’fragment of cDNA clone 121 were introduced into the PstI site of the plasmid Bluescript SK+ (18). The reconstituted full-length cDNA was excised with HindIII-BumHI enzymes and finally transferred to the PSV2-GPT vector (19) from which the GPT gene had been excised with HindIII-BumHI. This vector containing the entire coding sequence of preproapoA-I was identified as PSVB-preproapoA-I and was used to express human apoA-I in CHO cells. To prepare an expression vector for E. coli, the pGEX-2T vector developed by Smith and Johnson (9) was used. This vector is characterized by a tat promotor (20) followed by the complete sequence of glutathione S-transferase (Sj26) (21) in which a polylinker (BumHI, SmaI, and EcoRI sites) replaces the normal termination site of Sj26. This polylinker is preceded by the sequence encoding the amino acids recognized and cut by thrombin (Leu-Val-Pro-Arg-Gly-Ser). By cloning in this vector, it is possible to express, in E. coli, a desired cDNA as a fusion protein with glutathione S-transferase. However, if the coding sequence does not start with a BamHI site, the expressed protein will contain supplementary amino acids at its NH,-terminus. As this would be the case for human apoA-I, the following cloning was conducted. The cDNA clone 101 insert was excised by P&I, purified, and partially digested by the enzyme Suu3AI. Two oligonucleotides (oligo 1, 5’GATCCATCGAAGGTCGTGACGAGCCACCGCAGAGTCCATGG-3’; oligo 2,3’GTAGCTTCCAGCACTGCTCGGTGGCGTCTCAGGTACCCTAGB) were synthesized to obtain a double-stranded DNA fragment containing, from 5’to 3’, a BamHI adaptor sequence, 16 bp encoding the amino acid sequence recognized and cut by Factor Xa, 24 bp reconstituting the m issing sequence of the early apoA-I cDNA lost with Suu3AI digestion, and finally a Suu3AI adaptor sequence. This fragment was ligated to the Suu3AI-PstI fragment of the apoA-I cDNA together with a BumHI-PstI-cleaved Bluescript
8 2
200
2 P 2
too
3
i
0 2
4
6
8
10
12
Days in culture FIG. 1_. Effect of different culture media on the production of recombinant human apoA-I by CHO cells at different days of culture. At Day 0, apoA-I-CH04 cells (1.5 X 106) were seeded in lo-cm’ dishes containing 4 ml MEM supplemented with 5% FBS. At Day 2, the cells were approximately 90% confluent and the medium was replaced by MEM containing 5% FBS (I), 2% LPSR-I (f& ITS alone (O), or ITS with 1% FBS (8). Every other day, the production of recombinant human apoA-I was determined by solid-phase RIA and the medium was replaced.
plasmid (KS+). Positive clones were confirmed by sequencing and the insert was excised with BarnHI and EcoRI enzymes, purified, and introduced in a pGEX-2T plasmid digested with the same enzymes. The expression of this clone, identified as pGEX-BT-AI, together with a factor Xa digestion results in a fully mature apoA-I. Production and Purification of Recombinant upoA-I Produced in CHO Cells
Human
The presence of apoA-I in the culture medium of a pool of stable transformants of confluent CHO cells was estimated by solid-phase RIA at 0.1 t 0.05 pg/ml of culture medium (mean f SE, n = 4) for 2 days. High producer clones were isolated by lim iting dilution. After the screening of 480 clones, 3 producing 0.2 pg/ml and 1 producing 0.5 pg/ml of apoA-I were found. This cellular clone was designated apoA-I-CH04 and was used in the studies described below. Production of apoA-I was followed with different culture media to choose the best medium for production and purification. Figure 1 shows the concentration of human apoA-I in the medium at times ranging from 2 to 12 days of culture in DME-F12 medium containing either 5% FBS or 2% of a serum substitute (LPSR-I), or ITS (3 mg/liter bovine insulin, 3 mg/liter human transferrin, and 5 pg/liter sodium selenite) (22,23) alone or with 1% FBS. As shown, the production of apoA-I increases with longer periods of culture, with the exception of when ITS-supplemented medium is used. This medium induces a decline of production at Day 4 and
EXPRESSION TABLE Purification
of Recombinant Volume
Fraction Conditioned
medium
b&d
317
apoA-I
from
apA-
apOA-I yield (W)
bd
2250
CHO
apoA-I
299
Cells
Fold purification -
29 19
7OaO
162 107
100
Precipitation Phenyl-Sepharose
59
2
250
42
23
33
Affinity chromatography
1
96
6
415
10.4
Note. Recombinant human apoA-I was produced by apoA-I-CH04 cells cultured in MEM supplemented with 5% FBS and purified. Samples obtained at different steps of the purification were assayed for protein content by the Lowry method (15) and human apoA-I was determined by RIA (11).
from that day the production of human apoA-I is lower than that with the other media. The addition of 1% FBS to the ITS-supplemented medium improved production but it remained consistently lower than that with the medium containing 5% FBS. A similar production was also observed with the medium supplemented with 2% LPSR-1. The use of LPSR-1 was excluded since we found that this medium contains approximately as much bovine apoA-I as 5% FBS-supplemented medium. Thus, the medium containing 5% FBS was chosen for subsequent studies because of the level of production and the lack of an acceptable alternative. Purification of recombinant apoA-I from culture medium consisted of chromatography on phenyl-sepha-
A
HUMAN
1
Human Protein
(ml)
OF RECOMBINANT
B
1
234
1
12
5
FIG. 3.
Isoelectric focusing of recombinant human apoA-I produced by CHO cells and by E. coli. Recombinant human apoA-I was produced by apoA-I-CH04 cells and E. coli (pGEX-PT-AI) and the charges of these different preparations of apoA-I were analyzed by isoelectric focusing on a gel containing ampholytes from pH 4 to 6. At the completion of the electrophoresis, the proteins were transferred to nitrocellulose and apoA-I was detected with anti-human apoA-I mAb A05 (A), mAb 4Hl (B), or mAb 4A12 (C). (A) Lane 1, bovine serum; lanes 2 to 4, different preparations of recombinant human apoA-I produced by CHO cells; lane 5, human plasma apoA-I. (B) Lane 1, human plasma apoA-I; lane 2, recombinant apoA-I produced by E. coli. (C) Lane 1, recombinant apoA-I produced by E. coli. The arrows indicate the position of the mature form of apoA-I.
rose followed by immunoaffinity chromatography on an anti-apoA-I mAb 4Hl-Sepharose. This immunoabsorber was chosen because mAb 4Hl does not crossreact with bovine apoA-I (manuscript in preparation). This was also verified by RIA and by applying bovine serum onto a 4Hl immunoabsorber, which, respectively, resulted in no detection and no retention (data
44.2 27.0 18.3 -
12345
12
FIG. 2. Electrophoretic pattern of recombinant human apoA-I produced by CHO cells at different steps of the purification. Recombinant human apoA-I was produced by apoA-I-CH04 cells cultured in MEM supplemented with 5% FBS and purified. Samples obtained at different steps of the purification were NII on SDS-PAGE (12%) and either stained by Coomassie blue (A) or transferred to nitrocellulose and detected by a mixture of anti-human apoA-I mAbs (4H1, 5F6, 4A12) (B). (A) Lane 1, conditioned medium; lane 2, ammonium sulfate precipitate; lane 3, sample after phenyl-Sepharose chromatography; lane 4, sample after afEnity chromatography on mAb 4Hl-Sepharose; lane 5, human plasma apoA-I. (B) Lane 1, human plasma apoA-I; lane 2, purified recombinant human apoA-I.
123456
FIG. 4.
12
3
4
5
12345
Electrophoretic pattern of E. coli proteins at different steps of the purification. Recombinant human apoA-I was produced by E. coli (pGEX-2T-AI) and purified. The samples obtained at different steps of the purification were run on SDS-PAGE (12%) and either stained by Coomassie blue (A) or transferred to nitrocellulose and probed with anti-human apoA-I mAb 4Hl (B) or mAb 4A12 (C). (A, B, and C) Lane 1, E. coli cell extract; lane 2, cell extract purified by glutathione-Sepharose affinity chromatography; lane 3, purified extract digested by Factor Xa; lane 4, fraction of the digest not retained by glutathione-Sepharose affinity chromatography. (A) Lane 5, twice the amount of protein applied in lane 4; lane 6, purified human plasma apoA-I. (B and C) Lane 5, purified human plasma apoA-I.
300
BRISSETTE
not shown). The results of purification are summarized in Table 1 and the electrophoretic pattern of samples taken at different steps of the purification can be seen in Fig. 2A. The major protein found in the culture medium is a 65kDa protein presumed to be bovine albumin of the FBS (Fig. 2A, lane 1). As purification proceeds, a protein of 28 kDa appears (Fig. 2A, lane 3). This band comigrates with human serum apoA-I and coincides with a protein detectable on a Western blot with mAb anti-human apoA-I (Fig. 2B). Judged by the intensities of the band, the protein of 28 kDa represents more than 10% of the proteins purified by phenyl-Sepharose and more than 90% of the proteins detected by Coomassie blue staining of the SDS-PAGE, at the end of purification (Fig. 2A, lane 4). However, the contribution of apoA-I determined by RIA suggests that human apoA-I contributes to approximately 1 and 11% of the protein content of the preparations obtained after purification on phenyl-Sepharose and on mAb 4Hl-Sepharose, respectively (Table 1). This discrepancy between the two methods may be a consequence of the immunodetection of recombinant apoA-I produced by CHO cells being weaker than that of plasma apoA-I, as shown and discussed in the last part of the manuscript. The apoA-I is secreted as a molecule of proapoA-I which loses its prosegment of 6 amino acids to become a fully mature apoA-I (2). The proapoA-I usually represents less than 5% of plasma human apoA-I and its isoelectric point is 5.85 instead of 5.65 (24). Figure 3A compares the IEF patterns of three different preparations of purified recombinant human apoA-I with apoA-I purified from human plasma HDL. Thus, the preparations of recombinant apoA-I (lanes 2 to 4) are richer in a molecule with an isoelectric point comparable with human plasma proapoA-I (lane 5) than in the presumed mature form (approximately 30%). This experiment was conducted with mAb A05, which cross-reacts with bovine apoA-I (manuscript in preparation), but similar experiments with mAbs that do not recognize bovine apoA-I gave essentially the same results. The mAb A05 was chosen because bovine apoA-I was analyzed in the same experiment. Figure 3A also shows that the isoelectric point of bovine apoA-I (lane 1) is higher than that of the mature and the proform of recombinant human apoA-I. This corroborates the specificity of the 4Hl affinity column used for purification of human apoA-I, since no signal of an apoA-I molecule with this isoelectric point is detectable in the preparation of purified recombinant human apoA-I. Production and Purification of Recombinant Human apoA-I Produced by E. coli Approximately 20 mg of recombinant human apoA-I could be produced per liter of E. coli culture. Figure 4A shows an electrophoretic pattern of the proteins ob-
ET AL. TABLE 2 Purification
of Human Volume
Fraction Cell extract Affinity chromatography
(ml)
Recombinant
from E. coli
apoA-I
Protein
apoA-I
Wml)
Wml)
apoA-I yield (7%)
Fold purification
14
23,000
3ooo
100
-
20
!m
720
34
6.3
Note. Recombinant human apoA-I was produced by E. coli (pGEXST-AI) and purified. Samples obtained at different steps of the purification were assayed for protein content by the Lowry method (15) and human apoA-I was determined by RIA (11).
tained at different steps of the purification and Table 2 shows the extent of purification achieved. The glutathione-Sepharose affinity column retains a protein with an apparent molecular weight of 55 kDa, as expected for a glutathione S-transferase-apoA-I fusion protein (lane 2). Three other bands of molecular weight ranging from 30 to 35 kDa but with intensities lower than that of the 55-kDa protein are also observed. These preparations were assessed by immunoblotting with mAb 4H1, which recognizes an epitope close to the NH,-terminus (8), and mAb 4A12, which is specific for an epitope located in the last third of the apoA-I sequence (8). Figure 4C (lane 2) shows that mAb 4A12 recognizes a 55-kDa protein and a family of proteins of molecular weight ranging from 30 to 35 kDa, as well as some low-molecularweight peptides. Thus, the immunodetection of the purified fusion protein with mAb 4A12 and staining with Coomassie blue give similar patterns. This suggests that the family of proteins of size ranging from 30 to 35 kDa is composed of the first two-thirds of the apoA-I molecule linked to a terminal portion of the glutathione Stransferase molecule large enough to be retained by glutathione-Sepharose. A different pattern is obtained (Fig. 4B, lane 2) when a mAb recognizing the extreme NH,-terminal region of apoA-I is used for detection. Indeed, not only the bands detected with mAb 4A12 but also a family of proteins of 10 kDa and several less intense signals distributed between 55 and 10 kDa were observed. This implies that an important fraction of the fusion proteins contains the NH,-terminus of apoA-I but not the COOH-terminus. Furthermore, the more important contribution of these multiple bands on immunoblots than on Coomasie blue-stained gels suggests that the smaller fusion proteins are more immunoreactive and/or are less chromogenic than the full-size fusion protein. Several attempts were made to eliminate these multiple bands. Expression of apoA-I was determined at different times (0 to 3 h) after the induction with IPTG, different concentrations of IPTG were tested, and conventional protease inhibitors were added. Also, other
EXPRESSION
Bmmax
OF RECOMBINANT
4Hl
Bmmax IPJ
‘*el
Concentration
I
BIBmax
, 2 B/Bmax
3610
All
@g/nG
5F6 1.0 B’Bmax
I
,2 BfBmax
4A12
I
I
Concentration FIG. 5. Competitive protein was competed from E. coli (0).
301
apoA-I
(pg/ml)
2611
¢radon
HUMAN
@g/ml)
solid-phase radioimmunoassays. The interaction of different mAbs (4H1,2Gll, 3G10, All, 5F6, and 4A12) with HDL with purified human plasma apoA-I (A) or with purified recombinant human apoA-I obtained from CHO cells (Cl) or
procedures for protein extraction were tested (8) including one selecting for cytoplasmic and periplasmic proteins (25). E. coli POP-2136 that are protease deficient were also used as a host cell. Under all these conditions, there was no clear improvement (data not shown), suggesting that gene induction and proteolysis secondary to the breakage of the cells are probably not responsible for the smaller products detected. Consequently, this suggests that either transcription or translation is prematurely interrupted. Digestion of the fusion protein with Factor Xa gives a doublet of bands with an apparent molecular weight of 28 kDa on SDS-PAGE (Fig. 4A, lane 3). The smaller band comigrates with human apoA-I (lane 6) while the larger one is likely glutathione S-transferase as judged by its disappearance upon passage on glutathioneSepharose (lane 4). Several less intense bands of molec-
ular weight lower than 28 kDa could also be observed. In accordance with the results obtained with the fusion protein, immunodetection with mAb 4Hl revealed several bands in the digested sample while only bands of 28 kDa and of less than 10 kDa were detectable with mAb 4A12 (Figs. 4B and 4C, respectively, lanes 4). The isoelectric point of this recombinant human apoA-I was analyzed by immunodetection of Western blots of IEF gels. Figure 3C (lane 1) shows that mAb 4A12 detects a protein that comigrates with the mature form of plasma apoA-I. However, mAb 4Hl (Fig. 3B, lane 2) recognizes not only this apoA-I but also a series of molecules with lower isoelectric point. Together the immunodetection of Western blots of SDS-PAGE and IEF suggests that the intact apoA-I molecule produced by E. coli has the same isoelectric point as human plasma apoA-I.
302
BRISSETTE TABLE
3
Immunoreactivity of Human apoA-I Produced by CHO Cells and E. coli in Comparison with Purified Human Plasma apoA-I ED,
recombinant
mAb
CHO cells
4Hl 3G10 2Gll All 5F6 4Al2 N
2.2 3.4 6.7 2.1 7.6 5.5
f + f + f f
0.75” 1.6 2.1 0.05 2.4 0.85 2
apoA-I/ED, apoA-I
plasma
E. coli 0.12 3.7 8.5 0.08 8.7 7.6
f + f + + f
o.007b 1.4 4.3 0.009’ 3.6 0.88 3
Note. Recombinant human apoA-I obtained from CHO cells or E. coli was tested in competitive RIA with the mAbs indicated. ED, is the effective dose at 50% binding measured by RIA in the presence of Tween 20. a Values represent the mean ? SE. b P < 0.05. c P < 0.0001.
Comparison of the Expression of Epitopes between Recombinant Human apoA-I and Purified Human Plasma apoA-I
The characterization of the different recombinant human apoA-I was completed by an analysis of the expression of six epitopes distributed in the primary sequence of the apoA-I molecule. This was accomplished by competitive solid-phase RIA using different mAbs, including mAb 4H1, which recognizes an epitope mapped between amino acids 2 and 8; mAb 2Gl1, which is specific for an epitope determined between amino acids 44 and 112; and mAbs 2G10, All, 5F6, and 4A12, which recognize epitopes located between amino acids 98 and 121,99 and 132,118 and 148, and 173 and 205, respectively (8). Two preparations of human apoA-I obtained from CHO cells and three preparations from E. coli were tested with three preparations of purified human plasma apoA-I. The displacement curves for all rnAbs are shown in Fig. 5 for one preparation of each type of apoA-I and the mean and standard deviation of the ratios of the ED, (effective dose at 50% binding) of all preparations to the ED, of purified human plasma apoA-I appear in Table 3. Figure 5 shows that independent of the mAb used, the slopes of the competition curves are identical among the different preparations of apoA-I. This indicates that the aflinities of the different monoclonal antibodies for purified plasma apoA-I and recombinant human apoA-I are similar and suggests a similar conformation of the epitopes in the various apoA-I forms tested. However, for most RIA (2Gl1, 3Gl0, 5F6, 4A12), a shift in the curves toward higher
ET
AL.
apoA-I concentrations is observed with recombinant human apoA-I produced by either CHO cells or E. coli. Table 3 shows that the ED,, ratios determined with mAbs 2Gl1, 3G10, 5F6, and 4A12 are not significantly different between the recombinant human apoA-I forms. In these assays, a concentration of the two types of recombinant human apoA-I much higher (2.1- to 7.6fold) than that of purified plasma apoA-I is needed to achieve half the maximal binding with the mAbs. Thus, it can be hypothesized that either the contribution of recombinant apoA-I in the purified preparations has been overestimated or a fraction of the purified recombinant apoA-I has been modified in the process of purification. The second hypothesis must be rejected because we verified that the isolation of human plasma apoA-I by the methods used for CHO-derived apoA-I did not alter the immunoreactivity of apoA-I (data not shown). With E. coli-derived preparations, there is a net and significant shift toward apparent higher concentrations in apoA-I epitopes recognized by mAbs 4Hl and All (Fig. 5 and Table 3). This can be explained by the more abundant number of recombinant apoA-I molecules containing the NH,-terminus than the COOH-terminus and also by the higher immunoreactivity of smaller fragments of apoA-I. The reasons the epitopes of mAbs 2Gll and 3GlO mapped between the epitopes of mAbs 4Hl and All did not show a similar shift remain obscure. Thus, a different folding between human apoA-I produced in CHO cells and in E. coli cannot be excluded. CONCLUSION
In conclusion, we have produced and purified recombinant human apoA-I from procaryotic and eucaryotic systems and compared these apoA-I forms in terms of molecular weight, isoelectric point, and expression of several epitopes. Although the CHO stable transformant cells give a low yield of production, the recombinant apoA-I appears intact and its isoelectric point is compatible with that of the mature and the proform of apoA-I normally found in plasma. Despite the high yield of production of the E. coli system, part of the apoA-I produced is incomplete in the COOH-terminus. Furthermore, in contrast to the recombinant human apoA-I produced by CHO cells, two of six epitopes are expressed to a greater extent in apoA-I obtained from E. coli than from human plasma. Thus, the CHO system appears to be the best system to produce human recombinant apoA-I. ACKNOWLEDGMENTS We thank Dr. V. man apoA-I cDNA treal) for the CHO Milne for his useful We also gratefully
Zannis (Boston University) for providing the huand Dr. W. E. C. Bradley (Cancer Institute, Moncells. We are indebted to our colleague Dr. R. W. suggestions in the preparation of the manuscript. acknowledge the help of Ms. C. Lemire in typing
EXPRESSION
OF RECOMBINANT
the manuscript. This work was supported by the Medical Research Council of Canada (PG-2’7). L.B. is a recipient of a scholarship from the Canadian Heart Foundation.
REFERENCES 1. Swaney, J. B., Braithwaite, F., and Eder, H. A. (1977) Characterization of the apolipoproteins of rat plasma lipoproteins. Biochem-
i&y16,271-278. 2. Gordon, J. I., Smith, D. P., Andy, R., Alpers, D. H., Schonfeld,
3.
4.
5.
6.
7.
8.
9.
10. 11.
12.
G., and Strauss, A. W. (1982) The primary translation product of rat intestinal apolipoprotein mRNA is an unusual preproprotein. J. Biol. Chem. 257,971-978. Edelstein, C., Gordon, J. I., Toscas, K., Sims, H. F., Strauss, A. W., and Scanu, A. M. (1983) In vitro conversion of proapoprotein A-I to apoprotein A-I. J. Biol. Chem. 268, 11,430-11,433. Edelstein, C., Kaiser, M., Piras, G., and Scanu, A. M. (1988) Demonstration that the enzyme that converts precursor of apolipoprotein A-I to apolipoprotein A-I is secreted by the hepatocarcinoma cell line HepG2. Arch. Biochem. Biophys. 267,23-30. Belouchi, A., and Bradley, W. E. C. (1991) Analysis of second step mutation of class II and class III CHO APRT gene heterozygote: Chromosomal differences and dilution frequencies. Somatic Cell Mol. Genet., 17,277-286. Wigler, M., Pellicer, A., Silvertein, S., and Axel, R. (1978) Biochemical transfer of single-copy eucaryotic genes using total cellular DNA as donor. Cell 14, 725-731. Blocklinger, K., and Digglemann, H. (1984) Hygromycin B transferase as a stable marker for DNA transfer experiments with higher eucaryotic cells. Mol. Cell Biol. 4, 2929-2931. Marcel, Y. L., Provost, P. R., Koa, H., Raffai, E., Dac, N. V., Fruchart, J.-C., and Rassart, E. (1991) The epitopes of apolipoprotein A-I define distinct structural domains including a mobile middle region. J. Biol. Chem. 266,3644-3653. Smith, D. B., and Johnson, K. S. (1988) Single-step purification of polypeptides expressed in Escherichia coli as fusions with glutsthione S-transferase. Gene 6’7, 31-40. Tartof, K. D., and Hobbs, C. A. (1987) Improved media for growing plasmid and cosmid clones. Bethesda Res. Lab. Focus 9,12-13. Marcel, Y. L., Jewer, D., Vezina, C., Milthorp, P., and Weech, P, K. (1987) Expression of human apolipoprotein A-I epitopes in high density lipoproteins and in serum. J. Lipid Res. 28,768-777. Laemmli, U. K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227,680-685.
HUMAN
apoA-I
303
electro13. O’Farrell, P. H. (1975) High resolution two-dimensional phoresis of proteins. J. Biol. Chem. 250,4007-4021. 14. Marcel, Y. L., Jewer, D., Leblond, L., Weech, P. K., and Milne, R. W. (1989) Lipid peroxidation changes the expression of specific epitopes of apolipoprotein A-I. J. Biol. C&em. 264, 19,94219,950. 15. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) Protein measurement with the folin phenol reagent. J. Biol. Chem. 193,265-275. 16. Brewer, H. B., Ronan, R., Meng, M., and Bishop, C. (1986) Isolation and characterization of apolipoproteins A-I, A-II, and A-IV, in “Methods in Enzymology” (Segrest, J. P., and Albers, J. J., Eds.), Vol. 128, pp. 223-246, Academic Press, San Diego. S. K., Zannis, V. I., and Breslow, J. L. (1983) Isola17. Karathanasis, tion and characterization of the human apolipoprotein A-I gene. Proc. Natl. Acad. Sci. USA 80,6147-6151. 18. Short, J. M., Fernandez, J. M., Sorge, J. A., and Huse, W. D. (1988) Lambda ZAP: A bacteriophage lambda expression vector with in vivo excision properties. Nucleic Acids Res. 16, 7583-
7600. 19. Mulligan, R. C., and Berg, P. (1981) Selection for animal cells that express the Escherichia coli gene coding for xanthine-guanine phosphoribosyltransferase. Proc. Natl. Acad. Sci. USA 78,
2072-2076. 20. Amann, E., Brosius, J., and Ptashne, M. (1983) Vectors bearing a hybrid trplac promotor useful for regulated expression of cloned genes in Escherichia coli. Gene 25,167-178. 21. Smith, D. B., Davern, K. M., Board, P. G., Tiu, W. U., Garcia, E. G., and Mitchell, G. F. (1986) Mr 26 000 antigen of Schistosoma japonicum recognized by resistant WEHI 12915 mice is a parasite glutathione S-transferase. Proc. Natl. Acad. Sci. USA 83,97039707. 22. Hamilton, W. G., and Ham, R. G. (1977) Clonal growth of Chinese hamster cell lines in protein-free media. In Vitro 13, 537-
547.
23
Gasser, F., Mulsant, P., and Gillois, M. (1987) Long term multiplication of the Chinese hamster ovary (CHO) cell line in a serum-free medium. In Vitro Cell. Deu. Biol. 21, 588-592. 24 Zannis, V. I., Breslow, J. L., and Katz, A. J. (1980) Isoproteins of ’ human apolipoprotein A-I demonstrated in plasma and intestinal organ culture. J. Biol. Chem. 255,8612-8617. 25. Rozenwasser, T. A., Hogquist, K. A., Nothwehr, S. F., BradfordGoldberg, S., Olins, P. O., Chaplin, D. D., and Gordon, J. I. (1990) Compartmentalization of mammalian proteins produced in Escherichia coli. J. Biol. Chem. 265. 13,066-13,073. ’
