The expression, affinity purification and characterization of recombinant pseudomonas exotoxin 40 (pe40) secreted from escherichia coli

Biott-shnology ELSEVIER

Journal of Biotechnology 42 (1995)9-22

The expression, affinity purification and characterization of recombinant Pseudomonas exotoxin 40 (PE40) secreted from Escherichia coli John K. Kawooya a,1,Jeanne C. Treat a, Richard J. Kirschner a, Martha W. Sears a, Jonathan F. Gorczany b, Stephen H. Grode b, Diane S. Strother b, Paul A. Asmus ‘, Frances M. Eckenrode b!* a Bioprocess Research Preparations, The Upjohn Company, Kalamazoo, MI, 49001, USA b Analytical Methods and Services, The Upjohn Company, Kalamazoo, MI, 49001, USA ‘Analytical Development Biotechnology, The Upjohn Company, Kalamazoo, MI, 49001, USA

Received 27 October 1994; revised 30 March 1995; accepted 31 March 1995

Procedures have been devised for producing high yields of purified recombinant PE40, a protein important for the development of the anti-AIDS therapeutic, sCD4-PE40. PE40 is a truncated form of the bacterial toxin, Pseudomonas exotoxin A, it lacks the cell-binding domain, but retains domains II and III that are involved in membrane translocation and inhibition of protein synthesis in eukaryotic cells. Expression vectors in Escherichiu coli encoding PE40 synthesized the product as a soluble protein under control of the T7 promoter. The expression capabilities of transformants of E. coli BL21(DE3) were highly unstable. Expression levels (secreted and total) were evaluated in shake flasks and at the 10-l scale at 27°C and 37°C and following induction by IPTG or lactose. The cell-free media from the batch process was applied directly to a Cibacron blue F3GA-chromatographic medium and PE40 was eluted by nicotinamide in high yield and purity. This purification strategy was based on the structural similarity of the blue dye to NAD, a natural substrate for domain III of PE40. Green and red dye-ligand chromatography steps removed nicotinamide as well as minor residual E. coli proteins from PE40. Reversed-phase liquid chromatography peptide maps of purified PE40 were characterized by electrospray ionization mass spectrometry to determine the sequence and verify disulfide bonding. Keywords:

Pseudomonas exotoxin A; Expression optimization; ionization mass spectrometry

Cibacron blue F3GA; Peptide mapping; Electrospray

1. Introduction

* Corresponding author. ’ Present address: Cephalon, Inc., 145 Brandywine Parkway, West Chester, PA 19380-4245,USA. 0168-1656/95/$09.50 0 1995 SSDI 0168-1656(95)00055-O

Pseudomonas exotoxin A (PE) is a potent toxin secreted by Pseudomonas aeruginosa. PE inhibits protein synthesis in eukaryotic cells by catalyzing

Elsevier Science B.V. All rights reserved

1(I

.I.K. Ku wooyu Edal. /Journal

ADP ribosylation of elongation factor 2 (Iglewski and Kabat, 1975; Leppla, 1976). PE consists of three distinct domains which are defined according to their functions (Allured et al., 1986; Pastan et al., 1992). Domain 1 is the cell recognition domain which enables the toxin to bind to the cell receptor. Domain II facilitates the translocation of the toxin across the membrane of target cells into the cytosol (Hwang et al., 1987) and is important for secretion from E. coli (Chaudhary et al., 1988bl. Domain III possesses the catalytic site for ADP ribosylation (Hwang et al., 1987). A recombinant gene encoding only domains II and III was constructed (Chaudhary et al., 1988a; Kondo et al., 1988). PE40 is the 40 kDa recombinant version of PE lacking the cell-binding domain. Bacterial toxins are powerful tools in biomedical research when they are coupled with molecules which direct their toxicity towards specific cells. Recombinant chimerics have been constructed in which domain I of PE is replaced by cell-binding domains from different proteins (Fitzgerald, 19871. Such a chimeric protein was constructed by fusing sequences from CD4 with those of PE40 to generate sCD4-PE40 (Ashorn et al., 1990; Chaudhary et al., 1988al. In vitro, sCD4-PE40 selectively targets HIV-infected T4 lymphocytes through cell surface receptor gp120, and destroys them. Recombinant sCD4-PE40 was of interest to the Upjohn Company as a possible therapeutic for AIDS. The objective of the work described here was to provide purified PE40 to serve as a control molecule for analytical development and for clinical studies. Cultures and conditions were sought that would generate PE40 in the medium to simplify isolation and minimize the need for oxidation and refolding.

2. Materials and methods 2. I. Materials Yeast extract and tryptone were purchased from DIFCO, ampicillin from Bristol Alpha (Barcelloneta, P.R.), isopropyl-P-p-thiogalac-

of’ Biotechnology 42 (1995) 9-22

topyranoside (IPTG) from Boehringer Mannheim Biochemicals, adenosine S-diphosphate, p-nicotinamide adenine dinucleotide and dithiothreitol from Sigma, and trypsin (TPCK treated) from Worthington. Coomassie brilliant blue G-250 and Affi-gel blue dye-ligand chromatography media were purchased from Bio-Rad, mimetic green and mimetic red dye-ligand chromatography media from Affinity Chromatography, Ltd., UK. Bradford Protein Assay Reagent and trifluoroacetic acid from Pierce, and acetonitrile from American Burdick and Jackson. Spiral cartridges were purchased from Amicon. 2.2. Media All media described below incorporate M584 buffered salts: 15 mM K,HPO,, 50 mM Na(NH,)HP04.4H,O, 10 mM citric acid, and 5 mM (NH&SO,. The M584 Seed Medium for growing inocula also included 1 mM MgSO,, 1% yeast extract, and 0.8% glycerol. M584 Rich Medium included 10 g I-’ tryptone, 5 g I-’ yeast extract, 5 g 1-l NaCl, and glucose at either 0.5% for shake flask studies or 2.25% for mass cultures. M584 Minima1 Medium included 3 mM MgSO,, 10 mM oL-methionine, trace minerals 1.6 FM H,BO,, 120 nM (84 nM (NHJ,Mo,O,,, CoCl,, 40 nM CuSO,, 320 nM MnCl,, 40 nM ZnSO,, 75 FM FeSO, dissolved in 105 PM citric acid), 0.75 g 1-l antifoam (SAG 4130, Union Carbide), and 0.75% glucose. Ampicillin was added only to shake flask cultures subjected to passaging and to Seed Medium. IPTG (10 or 100 mM) and lactose (40%) were added to cultures from concentrated solutions. 2.3. Bacterial strains and plasmids E. coli strain BL21(DE3), obtained from F.W. Studier (Brookhaven National Laboratory), carries a chromosomal copy of the T7 polymerase gene under control of lacUV5; products were inducible by IPTG or lactose. The vectors pVC8 and pVC85 were constructed and supplied by V.J. Chaudhary of the National Cancer Institute, National Institutes of Health (Kondo et al., 1988). These are pBR322-based vectors and carry the

_I.K Kawooya et al. /Journal

PE40 gene under control of the T7 410 promoter; pVC85 also contains the ompA signal sequence to facilitate secretion of the product. These vectors were transformed into BL21(DE3) using standard procedures. One such transformation with pVC85 gave rise to KU183. Cultures were stored in the vapor phase of liquid nitrogen. 2.4. Passaging and expression evaluation in shake flasks Following transformation, BL21(DE3XpVC85) was diluted in Rich Medium containing 250 c.Lg ml-’ ampicillin and incubated overnight at 37°C. The cells were grown continuously, without chilling, by reinoculating into fresh medium at 103-lo5 ml-’ each time the cells reached 4 x 10’ ml-’ (about 1 A,,,). Clonal isolates (as well as the mixed population) from this extended passaging (100 generations) were screened for expression capability in shake flasks. Expression was evaluated by growing the cells in Rich Medium containing 250 pg ml-’ ampicillin at 37°C to a density of l-2 A,,,; 1 mM IPTG (final concentration) was then added, incubation at 37°C continued for 2 h post-induction, and samples were analyzed by SDS-PAGE. 2.5. Microbial growth Growth experiments were conducted in New Brunswick 14 liter MicroGen reactors equipped with dissolved oxygen and pH probes (Ingold) and were monitored and controlled by a proprietary software package. Operational parameters were: pH controlled at 6.9-7.0 with NH,OH and H*SO,; temperature set at 27°C or 37°C; air flow set at 1 wm (i.e., 10 slm air with 10 1 medium); backpressure set at 10 psi; agitation controlled at r 650 rpm; and dissolved oxygen controlled at > 40% with agitation. Inoculations were conducted at 0.01 A,,, using shake flask cultures grown in Seed Medium at 27°C or 37°C to a density of l-2 A,,,,. In the batch protocol, cells were grown to 10 A,,, in Rich Medium with 2.25% glucose before 0.2 mM IPTG (final concentration) was added to induce expression. In the fed-batch protocol, cells were

of Biotechnology 42 (1995) 9-22

11

grown in Minimal Medium containing 0.75% glucose, a sufficient carbon source to support growth only to lo-11 A,,,. The cells were glucose starved for a short period (30-60 min), then the 50% glucose feed was initiated at a rate of 8 ml 1-l h- ’ shortly before 6 mM lactose (final concentration) was added to induce the expression of product. These protocols were evaluated at both 27°C and 37°C. No antibiotics were added to the media. 2.6. Preparation of E. coli medium for chromatography

Spent media (about 10 1) were centrifuged to remove E. coli cells. Cell-free medium was concentrated to 500 ml using a 1 ft2 spiral ultrafiltration cartridge (Amicon), and equilibrated with 5 mM sodium phosphate, pH 6.0 at 4°C. 2.7. Blue dye-ligand chromatography The concentrated medium was applied to an Affi-gel blue dye-ligand column with a bed volume of 100 ml. During loading, the flow rate was maintained at about 2 bed volumes per h. Unbound protein was washed from the column with 5 mM sodium phosphate buffer, pH 6.0, until the ultraviolet absorbance of the eluate was baseline at 280 nm. The bound protein was eluted from the column in three fractions. The first fraction was eluted with 5 mM sodium phosphate/25 mM nicotinamide, pH 6.0. The column was then washed with 5 mM sodium phosphate buffer, pH 6.0, to remove the nicotinamide. The second protein fraction was eluted with 10 mM sodium borate/O.5 M guanidine-HC1/20% glycerol, pH 8.6. The third protein fraction was removed from the column with 10 mM sodium borate/6 M guanidine-HCl, pH 8.6. 2.8. Green dye-ligand chromatography The PE40 fraction from the blue dye-ligand column was concentrated and equilibrated with 20 mM sodium phosphate, pH 7.0 using a stirred ultrafiltration cell (400 ml, Amicon YMlO membrane). The solution was applied to a Mimetic

12

J.K. Kawooya et al. /Journal

green dye-ligand column (50 ml bed volume) equilibrated with the above buffer at 4°C. After washing the unbound material from the column, bound impurities were eluted with 10 mM sodium phosphate/O.3 M NaCl, pH 7.0. PE40 was then eluted from the column with a 0.35 to 1 M NaCl linear gradient (600 ml total volume). After the gradient step, the column was brought to room temperature and the residual bound protein was eluted with 10 mM sodium borate/O.5 M guanidine-HC1/25% glycerol, pH 8.6. 2.9. Red dye-l&and chromatography The PE40 fraction from the green dye-ligand chromatography step was equilibrated with 10 mM sodium phosphate, pH 6.0, and applied to a red dye-ligand column (50 ml bed volume) at 4°C. PE40 was eluted from the column by means of a 0.1 to 0.8 M NaCl linear gradient in 10 mM sodium phosphate, pH 7.0 (600 ml total volume). Fractions were analyzed by SDS-PAGE and those containing PE40 were pooled, concentrated, and dialyzed against 10 mM sodium phosphate, pH 7.0. This pool contained the final PE40 product from the three-step purification protocol. 2.10. SDS-polyacrylamide gel electrophoresis SDS-polyacrylamide gel electrophoresis was carried out on l-mm thick 10 to 20% polyacrylamide gels purchased from Integrated Separation Systems essentially as described by Laemmli (1970) and stained with Coomassie brilliant blue G-250. 2.11. Isoelectric focussing (1~87 IEF was performed in the absence of denaturant on l-mm thick PAGplate horizontal slab gels purchased from Pharmacia. Gel composition was 5% acrylamide, 3% crosslinker, pH range 3.5 to 9.5. pl standards were from Pharmacia. 2.12. Analysis of PE40 expression Samples from reactors or shake flasks were centrifuged for 5 min at 12000 X g. The cell-free

of Biotechnology 42 (1995) 9-22

supernatant was used to assess PE40 secreted into the medium and the pellet was used to determine cell-associated product. For shake flask samples, the cell-free fluid was treated with 0.25 volume of 50% trichloroacetic acid, chilled, and centrifuged to concentrate the product for analysis. PE40 in the periplasmic space was estimated from the release of PE40 into a cell-free fraction following a freeze-thaw treatment: the pellet from 10 Ass0 of cells was resuspended in 0.5 ml 10 mM Tris, pH 8, frozen on dry ice, thawed at 30°C centrifuged, and the supernatant collected. This was repeated a total of four times. In experiments not reported here, ‘periplasmic’ PE40 also was estimated using a chloroform shock protocol (Ames et al., 1984). Similar results were obtained by both methods. PE40 was determined by SDS-PAGE, and expressed either as titer (pg ml-‘) or specific expression level (% TCP, the amount of PE40 in the cell relative to the total cell protein). Stained and dried gels were scanned on a Molecular Dynamics Computing Densitometer (Model 300A). Quantitation was with respect to bovine serum albumin (l-2 pg) included on each gel as an external standard. The % TCP for the recombinant protein was estimated from the total titer (sum of the cell-associated and secreted titers) and the cell mass using the conversion factor 1 A 550= 0.26 mg dry weight per ml = 0.14 mg protein per ml. Cell mass was assessed by turbidity following dilution of the sample to 0.1-0.4 A,,, with 0.1 N HCl. 2.13. NAD glycohydrolase activity NAZI glycohydrolase activity was measured by an isocratic reversed-phase HPLC assay which quantitated liberated nicotinamide. The assay employed a Zorbax Rx C8 column (4.6 mm i.d. X 25 cm) eluted with 2% acetonitrile and 20 mM potassium phosphate buffer, pH 3.5. The flow rate was 1.5 ml min-’ with ultraviolet detection at 254 nm. NAD eluted at approx. 5 min and adenosine diphosphoribose and nicotinamide eluted at approx. 2.5 and 9 min, respectively. The reaction was initiated by the addition of 100 ~1 PE40 solutions (1 to 3 mg ml-’ total protein) to 1

.I.K Kawooyaet al. /Journal of Biotechnology42 (1995) 9-22

ml of 1.1 mM NAD in phosphate-buffered saline. Reactions were conducted at 30°C for 60 min, then quenched with 200 ~1 of 0.1 M phosphoric acid. 50 ~1 of the quenched reaction solution were injected for analysis. NAD glycohydrolase activity was determined as a percent of activity compared to a purified sCD4-PE40 standard. 2.14. Analytical reversed-phase HPLC Analytical RP-HPLC of PE40 was carried out on a Hewlett-Packard 1090 instrument equipped with a 4.6 x 250 mm Vydac Cl8 column. Gradient elution mobile phases were: (A) 0.1% aqueous trifluoroacetic acid (TFA); and (B) acetonitrile plus 0.085% TFA. The gradient elution was from 20% to 55% B in 35 min with a flow rate of 1 ml min-’ and ultraviolet detection at 215 nm. 2.15. Protein concentration The Bradford assay was used to determine protein concentrations of samples. Bovine serum albumin was used as an external standard protein. 2.16. Amino acid analysis Amino acid analysis was performed by the phenylthiocarbamyl (PTC) method (Heinrikson and Meredith, 1984), essentially as described previously (Bidlingmeyer et al., 1984). 2.17. Trypsin digestion The digestion mixture included 2.5 nmol PE40 (100 ~1 of a 1 mg ml-’ solution in phosphatebuffered saline), 100 ~1 of 0.1 M (NH,),CO,, pH 8.4, and 1.7 pg trypsin (a 1:60 weight ratio using 0.25 mg ml-’ trypsin). Reduced and non-reduced digestions were conducted overnight at room temperature. For reduced maps, a 50-fold molar excess of dithiothreitol (10 mg ml-’ solution) over cysteine residues was added, and the digestions were incubated an additional 2 h at 37°C. Digestions were quenched by adding TFA to a final concentration of 1% (volume).

13

2.18. Peptide mapping

Peptide mapping was conducted by RP-HPLC on a Varian 1090 liquid chromatograph in conjunction with electrospray ionization mass spectral analysis. The column was a Vydac Cl8 (0.46 X 25 cm) eluted at a flow of 1 ml min- ’ with detection at 215 nm. Gradient elution mobile phases were: (A) 0.1% aqueous (TFA); and (B) 10% water/90% acetonitrile plus 0.085% TFA. Gradient elution was as follows: 0 to 18% B in 22 min; 18 to 25% B in 20 min; and 25 to 56% B in 28 min. 2.19. Electrospray ionization mass spectrometry Reversed-phase liquid chromatography mass spectrometry data were collected and processed on a Finnigan TSQ-70 triple stage quadrupole mass spectrometer equipped with a thermospray-electrospray adaptor made in-house. The eluting tryptic fragments were electrospray ionized at a needle voltage of 3500 and a bias voltage of 50. The probe, collar, and ion source temperatures were set to 210°C 45°C and 200°C respectively. The mass spectrometer was programmed to pass ions through Ql and Q2, while scanning Q3 from 250 to 2000 amu in 2 s. The resolution was opened, and the calibration adjusted, such that the observed ion corresponded to the average molecular weight of the tryptic fragment. The mobile phase composition to the mass spectrometer was kept at a constant 50% aqueous by use of a second HPLC pump operated at a flow of 1 ml min- ’ which delivered a gradient that exactly mirrored that of the chromatography pump. A column was placed in-line to approximate the back pressure and dead volume found in the analytical HPLC. The split flow at the needle was adjusted to deliver approx. 5 ~1 min-’ to the mass spectrometer.

3. Results 3.1. Expression strain development An expression vector encoding PE40 (Kondo et al., 1988) was obtained from V.J. Chaudhary

J.K. Kawooya et al. /Journal

14

and I. Pastan (National Cancer Institutes, National Institutes of Health) in order to produce PE40. To simplify production of recombinant PE40, a strain with the following characteristics was desired: chemically-induced expression, secretion of catalytically active product into the medium at > 100 pg ml-‘, and stability during storage. Strains that would induce PE40 following 27*C

n

0

;

l

0

T

_

!secmted

-

7

1 Time

Pmiplaamic

2

0

I

t

3

4

(Hours Post-induction)

37OC

0

1

Time

2

3

4

(Hours Post-lnductlon)

Fig. 1. PE40 expression by strain KU183 as a function of time and temperature. KU183 was grown in shake flasks containing MS84 Rich Medium and induced with IPTG.

of Biotechnology 42 (1995) 9-22

addition of IPTG or lactose were generated by using expression vectors which place PE40 under control of the T7 late promoter, 410 (Kondo et al., 1988; Iglewski and Kabat, 197.51, and the strain E. co/i BL21(DE3), which carries the T7 RNA polymerase gene under the control of the lacUV5 promoter. PE is normally produced by its Gram-negative parent as an extracellular protein from which a hydrophobic leader sequence has been removed (Gray et al., 1984). Domain II, which is required for membrane translocation into target cells, has been shown to be important for secretion of PE40 from E. coZi (Chaudhary et al., 1988b). Accordingly, expression was determined from pVC8 which links the PE40 sequence with its putative internal secretion signal directly to the T7 regulatory region. In addition, expression was assessed from pVC85 which supports secretion by fusing the ornpA signal sequence to the PE40 gene. To avoid potential concerns about strain stability, freshly transformed cells that had not been stored or chilled were inoculated into shake flasks containing Rich Medium, grown at either 27°C or 37°C and induced with IPTG as described in Materials and methods. PE40 represented 5--6% of the total cell protein (%TCP) with strain BL21(DE3XpVC8), however, little or no product was evident in the cell-free medium. BL21(DE3XpVC85) expressed lo-15% TCP as PE40. Although the total expression levels for these two strains were essentially independent of temperature, the amount secreted from BL21(DE3)(pVC85) was temperature dependent; 3-4% of the PE40 was secreted at 27°C and 20-25% was secreted at 37°C. Because of the higher total and secreted expression levels, subsequent studies used BL21(DE3XpVC85) transformants. Based on experience with sCD4-PE40 compared to E. coli expression strains for soluble CD4 and other heterologous proteins, it was expected that a strain expressing the PE40 moiety might be unstable. Accordingly, a ‘passaging’ approach was implemented for the BL21(DE3XpVC85) transformants. The transformant population was grown for 100 generations at low cell density under favorable conditions of

.I.K Kawooya et al. /Journal

temperature and nutrition. An aliquot of the exponentially growing, passaged culture was frozen overnight at - 190°C in 20% glycerol to select against cells that might be unstable to storage conditions. These cells were thawed, passaged for another 25 generations, and plated on nutrient agar. Twenty-five isolates from the passaged population were screened for expression potential at the shake flask level; they differed little in growth, expression potential W-20% TCP as PE40), or secretion level (about 20% secreted to the medium). Expression strain KU183 was one of these isolates. PE40 expression by strain KU183 was evaluated in shake flasks as a function of time at both 27°C and 37°C; Fig. 1 shows the distribution of product among the cell compartments. At 27°C total expression reached 90 Fg ml- ’ by 3 h post-induction with < 5% secreted to the medium, but about 50% accumulated in the periplasmic space. Cells grown at 37°C expressed PE40 at the same total level (90 pg ml-‘), however, approx. 20% was secreted into the medium with a similar amount found in the periplasm. This temperature dependence of secretion is consistent with results found for unpassaged transformants.

of Biotechnology 42 (1995) 9-22

0’

0

I

I

I

I

I

1

2

4

6

8

to

12

Stomge Time (Months at -190C) Fig. 2. The effect of storage time at - 190°C on the expression capability of KU253. Samples were thawed after the indicated storage time and the amount of PE40 expressed was determined in shake flasks.

ml-‘) during the first 3 months of storage and by 80% (29 pg ml-‘) after 12 months. The proportion of secreted product dropped from 36% (0 time) to 20% at 6 months and remained stable thereafter (Fig. 2).

3.2. Strain stabilio

3.3. Optimization of growth and PE40 expression

Following 18 months of liquid nitrogen storage, the PE40 expression strain, KU183, expressed relatively little product; this PE40 was difficult to isolate and contained unusually high levels of E. coli protein contamination. A sample of the twice-passaged original transformation mixture, which had been retained in liquid nitrogen, was subjected to a third passaging and isolates were screened for expression capability. A single isolate was chosen as a replacement for KU183 and designated KU253. Expression in shake flasks by KU253 at 37°C was 110 mg l- ‘, with > 30% secreted into the medium and was superior to expression by the KU183 recovered from storage. PE40 expression capability of KU253 stored in liquid nitrogen was assessed as a function of time. Expression declined nearly 50% (140 to 75 pg

Several growth protocols were compared at the 10-l scale with KU183 (prior to long-term storage) including batch runs with both rich and defined (minimal) medium, and (glucose) fedbatch runs in defined medium at 27°C and 37°C. The batch runs were induced by the addition of IPTG, whereas fed-batch runs were induced with lactose. Two other protocols, permissive for rapid cell metabolism, appeared to severely stress the PE40 expression strain. A rich medium batch protocol at 27°C did not support good growth or expression; cell mass was maximum at about 8 A 550 and PE40 expression totaled only 200 mg 1-l. A high level of physiological stress, as indicated by extensive cell lysis, was apparent using a minimal medium fed-batch protocol at 37°C; cell mass peaked at 22 A,,, and PE40 expression totaled 470 mg 1-l before cells began to lyse.

J.K. Kawooya ei al. /Journal

16

Secretad

-

of Biotechnology 42 (199.5) 9-22

chromatography approaches analogous to those used for the NAD-binding lactate dehydrogenase (Ryan and Vestling, 1974) were explored to improve the yield and purity of the PE40. The approach was based on the specificity of PE40 domain III for NAD; PE40 catalyzes the transfer of the ADP-ribose moiety of oxidized NAD to a modified histidine of protein synthesis elongation factor 2 (Allured et al., 1986; Lory and Collier, 1980).

Batch

3.5. Blue dye-ligand chromatography -0

I

2

Time (Hours

3

4

5

6

Post-Induction)

Fig. 3. Comparison of PE40 expression by KU253 at the 10-I scale in batch and fed-batch processes conducted at 27°C.

The batch and fed-batch protocols employing minimal medium at 27°C resulted in relatively high cell density (24-30 A,,,) and high levels of PE40 expression. Fig. 3 demonstrates differences in production characteristics between the 27°C batch and fed-batch protocols at the 10-l scale with KU2.53. The fed-batch protocol provides higher total titers for PE40 (830 mg I- ’ vs. 640 mg 1-l) but the batch protocol results in greater secretion (36% vs. 18%) over a shorter expression period (4 h compared to 6.5 h). In both protocols, the cells displayed a low level of unregulated expression, as indicated by the presence of PE40 at induction in both the cells and medium. Accordingly, the batch protocol at 27°C was usually used to provide crude PE40 for downstream purification. 3.4. Purification of

Cibacron blue F3GA was used as the affinity ligand because of the structural similarity between nicotinamide and the terminal sulfonated benzene ring of this blue dye-ligand (Burton et al., 1990; Drocourt et al., 1978). Although dyeligands have the capacity to bind certain proteins in a specific manner, such specificity may be abolished by the effects of temperature, pH, and ionic strength (Scopes, 1986). In general, high salt concentrations and elevated pH and/or temperature tend to diminish protein-ligand interactions, while reverse conditions promote stronger interactions. Conditions for binding PE40 to the blue dye-ligand matrix were selected with the objective

-B7

PE40

The published protocol for purifying PE40 from crude E. coli medium was based on a combination of ion exchange and size exclusion chromatography steps (Kondo et al., 1988). Although we were able to purify PE40 according to this protocol, less than 20% of the PE40 was recovered. The high losses were attributed to poor recovery at each chromatography step. Affinity

Fig. 4. Elution profile of protein fractions collected during blue dye-ligand chromatography. Detector response was measured at 280 nm. Peak A, voided fraction; peak B, PE40 fraction eluted with nicotinamide (nicotinamide contributed to the amplitude of this peak); peak C, fraction eluted with 0.6 M guanidine_HC1/25% glycerol; peak D, protein fraction eluted with 0.6 M guanidine-HCI.

J.K. Kawooya et al. /Journal

of Biotechnology 42 (1995) 9-22

of minimizing non-specific (ionic and hydrophobic) interactions between ligand and protein. Optimum PE40 binding conditions included pH 6, low ionic strength and ambient temperatures. At pH 6, PE40 had a net negative charge; this was verified by the inability of PE40 to bind Q-Sepharose under the same conditions. Therefore, it is unlikely that the binding of PE40 to the dye-ligand was due to ionic interactions. This conclusion is supported by the observation that PE40 was readily eluted from the column by the addition of a low concentration of nicotinamide to the loading buffer, without altering the pH or ionic strength. The profile depicting the dye-ligand elution steps is presented in Fig, 4. SDS-PAGE and RP-HPLC analyses of the various components from this chromatographic step showed that the crude cell-free medium contained a variety of proteins that included about 5% PE40 (Fig. 5,

Mr=l x 10m3

1

2

3

4

5

6

17

i E

7

t 0

I

I

5

10

I

15

I

20

Time (minutes) Fig. 6. RP-HPLC analysis of PE40 fractions eluted from the blue dye-ligand affinity column: (A) culture medium, (B) voided fraction, (C) nicotinamide-eluted fraction, (D) fraction eluted with 0.6 M guanidine/25% glycerol, (E) 6 M guanidine-eluted fraction, (F) PE40 final product obtained by subjecting fraction in C to a combination of green and red dye-ligand chromatographies.

Fig. 5. SDS-PAGE analysis of PE40 fractions eluted from the blue dye-ligand affinity column. Lane 1, molecular weight standards; lane 2, culture medium; lane 3, voided fraction; lane 4, nicotinamide-eluted fraction; lane 5, fraction eluted with 0.5 M guanidine/25% glycerol; lane 6, 6 M guanidineeluted fraction; lane 7, PE40 final product obtained by subjecting fraction in lane 4 to a combination of green and red dye-ligand chromatographies.

lane 2 and Fig. 6, chromatogram A). Upon loading this material onto the blue dye-ligand column, a large peak of ultraviolet absorbance at 280 nm was present in the void volume (Fig. 4, peak A). The voided material contained an array of E; colt’ proteins in addition to PE40 (Fig. 5, lane 3 and Fig. 6, chromatogram B). Addition of nicotinamide to the loading buffer resulted in elution of another major peak (Fig. 4, peak B). The amplitude of this peak was attributed to the strong absorbance characteristics of nicotinamide at 280 nm. Analysis of the material in this peak revealed a single major constituent (Fig. 5, lane 4 and Fig. 6, chromatogram C) subsequently identified as

J. K. Kawooya et al. /Journal

18

of Biotechnology 42 (1995) 9-22

Table 1 Comparison of the amino acid composition deduced from the cDNA with that measured by analysis of a purified PE40 hydrolysate

rities were eliminated from PE40 by a combination of green and red dye-ligand column chromatographic steps.

Residue

Measured

Deduced

Aspartic acid Glutamic acid Serine Glycine Histidine Arginine Threonine Alanink Proline Tyrosine Valine Methionine Isoleucine Leucine Phenylalanine Lysine

29.8 53.6 18.3 37.3 5.6 28.6 15.9 42.8 29.2 10.8 18.0 0 12.9 41.3 9.2 3.5

29 48 19 39 6 30 17 48 26 11 20 0 13 40 9 3

3.6. Green dye-ligand and red dye-ligand chromatographies The purity of PE40 was further enhanced by passage over green and red dye-ligand columns. When applied to the green or red dye-ligand columns, PE40 bound to the ligands under the conditions described in Materials and methods. Residual nicotinamide was voided from the green ligand column. PE40 was eluted from the column with a linear gradient of 0.35 to 1 M sodium chloride. The PE40 fraction from the green dye-

’

PI

2

9.3 PE40. Based on SDS-PAGE and RP-HPLC analyses, the purity of PE40 was increased from 5% to 90% based on this single chromatography step. To further examine the specific interaction between PE40 and the blue dye-ligand matrix, the protein remaining bound to the column was eluted in a step-wise manner by means of buffers containing different concentrations of guanidine HCI at a higher pH. Fig. 5, lane 5 and Fig. 6, chromatogram D show protein that was retained on the column after eluting PE40 with nicotinamide. These were eluted from the column with buffer containing 0.5 M guanidine HCI and 25% glycerol. Fig. 5, lane 6 and Fig. 6, chromatogram E show the remaining protein stripped from the column with 6 M guanidine HCl. The ratio of A 260 to A28O measured in this fraction was 1.5: 1.0, suggesting the presence of nucleic acids. The absence of detectable PE40 in both these fractions demonstrated that virtually all PE40 bound to the dye-ligand was recovered in a single step. Despite the high degree of specificity with which PE40 was eluted from the column with nicotinamide, post-blue dye-ligand material contained minor E. cofi protein contaminants. These impu-

8.7 8.5 8.2

7.4 6.9 6.6

5.9 5.2 4.6 3.5 L Fig. 7. Isoelectric focusing analysis of purified PE40. Lane Pharmacia pZ standards; lane 2, purified PE40.

1,

J.K Kawooya et al. /Journal

of Biotechnology 42 (1995) 9-22

19

ligand column was bound to the red dye-ligand column and eluted by means of a linear gradient of 0.1 to 0.8 M sodium chloride. After green dye and red dye chromatography steps, purity of PE40 was enhanced from 90% to about 98% (Fig. 5, lane 7 and Fig. 6, chromatogram F). The recovery after these three dye-ligand chromatography steps were applied to the cell-free medium was about 60%. 3.7. Biological activity NAD serves as a substrate for the ADP-ribosylation activity of PE40, and NAD glycohydrolase activity (Anderson and Yost, 1986) is a convenient measure of ADP riboslylation (Lory and Collier, 1980). The NAD glycohydrolase activity of PE40 was determined by measuring the liberation of nicotinamide from NAD using an isocratic RP-HPLC assay. When crude E. coli medium was subjected to blue dye-ligand chromatography,

Time (minutes)

Fig. 9. Comparison of reduced (inverted) and non-reduced tryptic maps of PE40. Absorbance was measured at 215 nm. Tryptic fragments are labelled according to their position within the PE40 sequence. Disulfide bonded peptides are indicated by a plus sign, T1+4 indicates that Tl and T4 are bound by the 19-41 disulfide bond. Fragments not digested at predicted Arg and Lys residues are indicated by a dash. Only those peaks unique to the reduced map are labelled.

the NAD glycohydrolase activity in the fraction increased g-fold. This value is consistent with the concomitant enhancement in the purity of PE40 from 5% to 90%. Part of the NAD glycohydrolase activity in the crude sample may have resulted from NAD glycohydrolases of E. coli origin. 3.8. Characterization of purified PE40

Fig. 8. The amino acid sequence and predicted tryptic fragments of PE40. The sequence is predicted from the sequence of cDNA. Two disulfide bonds that bridge positions 19-41 and 126-133 are indicated by solid lines.

The cDNA sequence of PE40 predicts 367 amino acid residues with a calculated mass of 39396 kDa (Kondo et al., 1988). The experimentally determined amino acid sequence of the 20 N-terminal amino acids of PE40 isolated by the procedure described herein was identical to the predicted sequence. No N-terminal methionine was detected. The amino acid composition shown in Table 1 agreed well with that of expected values. PE40 has a theoretical pZ of 4.7, assuming unperturbed pK values for individual amino acids. IEF analysis of purified PE40 (Fig. 7) showed a major band migrating at a pZ slightly more basic than the 4.6 standard. Minor bands were observed at higher and lower pZ values.

20

J.K Kawooya et al. /Journal

3.9. Tryptic peptide mapping

The tryptic peptide map of isolated PE40 was characterized by liquid chromatography-electroTable 2 Tryptic peptides characterized mass spectrometry

by electrospray

ionization

A. Non-reduced map Tryptic Amino acid peptide residues 1+4 2-3 3 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 27-28 29 30-32 30-33 33

l-28 + 34-47 29-33 31-33 48-56 57-67 68-84 85-91 92-113 114-116 117-146 147-166 167-175 176-181 182-186 187-210 211-212 213-221 222-244 245-246 247-248 249-254 255-259 260-267 268-283 284-292 293-305 306-317 293-317 318-330 331-363 331-367 364-367

Retention Theoretical time (min) mol. wt. a

Observed ion h

57.2 4612.3 12.4 692.8 not found 399.5 47.6 989.2 52.5 1357.5 35.2 1584.7 15.2 856.9 42.3 1244.4 18.1 420.5 53.4 3772.0 49.5 2090.2 28.4 1090.2 21.7 736.9 16.5 673.7 65.2 2575.0 not found 245.3 43.5 1103.2 52.2 2488.7 not found 231.3 9.2 287.4 25.1 642.8 24.8 632.8 36.3 926.1 37.7 1614.8 37.7 1015.3 28.4 1343.4 57.8 1397.7 60.1 2723.1 39.9 1365.6 48.8 3420.8 49.5 3906.3 not found 503.6

1538.5 + + + 693.3 990.5 679.9 + + 793.3 + + 857.7 623.4 t t 421.4 1257.7 t t 1046.5 + t 546.2 + + 737.5 674.6 1288.1 t t

+

552.7tt

t t

B. Reduced map Ttyptic 1 4 11

Amino acid residues

Retention Theoretical time (mitt) Mol. Wt. a

peptide ion ’

l-28 34-47 117-146

58.2 45.5 53.0

1462.2 t + 847.7 t + 1258.7 t t

2922.3 1692.0 3774.1

spray ionization mass spectrometry to confirm the predicted amino acid sequence and disulfide bonding, as shown in Fig. 8. Reduced and non-reduced tryptic maps of PE40, reflecting ultraviolet detection at 215 nm, are shown in Fig. 9. Peaks in the figures are labelled according to the corresponding tryptic peptides as identified by electrospray mass spectrometry. Table 2 presents mass spectral data for identified peaks in the non-reduced map and unique peaks in the reduced map. Included are predicted tryptic peptide sequences, observed retention times, theoretical masses and observed ions. Observed ions represent singly, doubly or triply charged ions and are consistent with theoretical masses. Of the 33 predicted peptides, evidence was found for 30. Those not observed include two dimers, T17 and T21, and a trimer, T3. These small fragments likely elute close to the mobile phase front. Several peaks were identified as peptides resulting from incomplete digestion at predicted trypsin cleavage sites including T2-3, T27-28, T30-32 and T30-33. 3.10. Disulfide bonding

1245.2 + + 288.0 643.7 633.6 927.0 808.8 + + 508.6 t t 672.6 t t 699.8 t t 1362.4 t t 684.0 t t 1141.5 t + 1303.2+ t

of Biotechnology 42 (1995) 9-22

Observed

PE40 theoretically contains two disulfide bonds linking cysteine residues 19 and 41 (Tl and T4) and cysteine residues 126 and 133 (both in Tll). An ion corresponding to disulfide linked fragments Tl and T4 was observed at 57.2 min in the non-reduced map. In the reduced map, two peaks were observed at 58.2 min and 45.5 min for Tl and T4, respectively. An ion for Tll containing an internal disulfide bond was observed eluting at 53.4 min in the non-reduced map. In the reduced map, an ion of higher mass was observed at nearly the same retention time, corresponding to the reduced Tll fragment. Thus, the four cysteines are disulfide bonded sequentially, as was demonstrated by X-ray crystallography of full length PE (Allured et al., 1986).

t

a Calculated masses were determined using the following atomic weights: C = 12.011, H = 1.008, N = 14.007, 0 = 15.999 and S = 32.060. b The ions listed correspond to M t H unless designated with + + or + + + , which indicate doubly and triply charged ions, respectively.

4. Discussion A strain, KU183, which expressed relatively high levels of PE40 was isolated from a transfor-

J.K. Kawooya et al. /Journal

mation mixture following passaging under conditions that selected against unregulated expression. However, long-term stability of PE40 expression strains, even when stored at - 190°C was difficult to achieve. This was demonstrated by strain KU253, a parallel clonal isolate of KU183, which lost approx. 50% expression capability after 90 d storage and approx. 80% after 12 months. Growth conditions were defined that resulted in expression of greater than 600 pg ml-’ PE40 with approximately 35% secreted. Both batch and fed-batch protocols yielded high levels of PE40, although the batch process was superior in the amount of secreted product and was preferable for generating product for isolation. The binding of PE40 to Cibacron blue F3GA is reminiscent of the interaction between this dye-ligand and some dehydrogenases (Biellman et al., 1979). In these examples, X-ray crystallography data revealed nucleotide binding sites that bound the dye-ligand. These interactions are thought to be based on the structural similarity between nicotinamide and the terminal sulfonated benzene ring of Cibacron blue F3GA. This mimicry was exploited to purify alcohol dehydrogenase by eluting the enzyme from Cibacron F3GA with low concentrations of nicotinamide (Ryan and Vestling, 1974). Likewise, these similarities were considered as a basis for using nicotinamide to elute PE40 from the dye-ligand in a selective manner. The NAD-binding pocket of the catalytic subunit of PE40 (domain III> is presumed to be the binding site for Cibacron blue F3GA. PE40 from crude E. coli medium was eluted from the blue dye-ligand column when low concentrations of nicotinamide were added to the elution buffer. Using these conditions, the purity of PE40 was enhanced from 5% in the crude medium to 90% after blue dye-ligand chromatography; the NAD glycohydrolase activity increased 8-fold. Green and red dye-ligand chromatography steps were added to the purification scheme to eliminate the nicotinamide that carried over from the blue dye-ligand chromatography step and to eliminate residual E. coli proteins and nucleic acids. Subsequent passage over green and red dye-ligand columns increased the purity to about

of Biotechnology 42 (1995) 9-22

21

98%, as determined by SDS-PAGE and RPHPLC. The overall recovery of PE40 from the three column purification procedure was about 60%. Amino acid analysis of the purified PE40 was consistent with that deduced from the cDNA sequence of the protein. IEF revealed a pl for the protein close to the calculated value of 4.6. Reduced and non-reduced tryptic peptide maps characterized by high performance liquid chromatography-electrospray mass spectrometry provided evidence for 30 of the 33 predicted tryptic fragments and verified the predicted linear disulfide bonding pattern.

Acknowledgements The PE40 gene and expression vectors were constructed in the laboratory of Dr. I. Pastan; we are grateful to Drs. V.J. Chaudhary and I. Pastan for making these available to us and for helpful discussions. We thank W.T. Caldwell for production support, and C.M. Campbell and Dr. J.G. Hoogerheide for performing N-terminal sequence analysis and amino acid analysis, and S.K. Henegar and B.J. Pechota for running SDS-PAGE and IEF. The TSP-ES1 was designed and constructed at the Upjohn Company by R.H. Robins and B. Stiemsma. We thank Dr. J.J. Manis for critically reviewing the manuscript.

References Allured, VS., Collier, R.J., Carroll, SF. and McKay, D.B. (1986) Structure of exotoxin A of Pseudomonas aeruginosa at 3.0~Angstrom resolution. Proc. Natl. Acad. Sci. USA 83, 1320-1324. Ames, G.F.-L., Prody, C. and Kustu, S. (1984) Simple, rapid, and quantitative release of periplasmic proteins by chloroform. J. Bacterial. 160, 1181-1183. Anderson, B.M. and Yost, D.A. (1986) Spectrophotometrtc assay of NADase-catalyzed reactions. Methods Enzymol. 122, 169-173. Ashom, P., Moss, B., Weinstein, J.N., Chaudhary, V.K., Fitzgerald, D.J., Pastan, I. and Berger, E.A. (1990) Elimination of infectious human immunodeficiency virus from human T-cell cultures by synergistic action of CD4-Pseudomonas exotoxin and reverse transcriptase inhibitors. Proc. Natl. Acad. Sci. USA 87, 8889-8893.

22

J.K. Kawooya et al. /Journal

Bidlingmeyer, B.A., Cohen, S.A. and Tarvin, T.L. (1984) Rapid analysis of amino acids using pre-column derivatization. J. Chromatogr. 336, 93-104. Biellman, J.F., Samama, J.P., Branden, C.I. and Eklund, H. (1979) X-ray studies of the binding of Cibacron blue F3GA to liver alcohol dehydrogenase. Eur. J. Biochem. 102, 107-110. Burton, S.J., Stead, C.V. and Lowe, C.R. (1990) Design and applications of biomimetic anthraquinone dyes III. Anthraquinone-immobilised C.I. Reactive Blue 2 analogues and their interaction with horse liver alcohol dehydrogenase and their adenine nucleotide-binding proteins. J. Chromatogr. 508, 109-125. Chaudhary, V.K., Mizukami, T., Fuerst, T.R., Fitzgerald, D.J., Moss, B., Pastan, I. and Berger, E.A. (1988a) Selective killing of HIV-infected cells by recombinant human CD4Pseudomonas exotoxin hybrid protein. Nature 335, 369372. Chaudhary, V., Xu, Y.-H.. Fitzgerald, D. Adhya, S. and Pastan, I. (1988b) Role of domain II of Pseudomonas exotoxin in the secretion of proteins into the periplasm and medium by Escherichia cob. Proc. Nat]. Acad. Sci. USA 85, 2939-2943. Drocourt, J-L., Thang, D.-C. and Thang, M-N. (1978) Blue dextran sepharose affinity chromatography: recognition of a polynucleotide binding site of a protein. Eur. J. Biochem. 82,355-362. Fitzgerald, D.J. (1987) Construction of immunotoxins using Pseudomonas exotoxin A. Methods Enzymol. 151, 139-145. Gray, G.L., Smith, D.H., Baldridge, J.S., Harkins, R.N., Vasil, M.L., Chen, E.Y. and Heyneker, H.L. (1984) Cloning, nucleotide sequence, and expression in Escherichiu cofi of the exotoxin a structural gene of Pseudomonas aeruginosa. Proc. Natl. Acad. Sci. USA 81, 2645-2649.

of Biotechnology 42 (1995) 9-22 Heinrikson, R.L. and Meredith, SC. (1984) Amino acid analysis by reverse-phase high-performance liquid chromatography: precolumn derivatization with phenylisothiocyanate. Anal. Biochem. 136, 65-74. Hwang, J., Fitzgerald, D.J., Adhya, S. and Pastan, I. (1987) Functional domains of Pseudomonas exotoxin identified by deletion analysis of the gene expressed in E. coli. Cell 48, 129-136. Iglewski, B.H. and Kabat, D. (1975) NAD-dependent inhibition of protein synthesis by Pseudomonas aeruginosa toxin. Proc. Nat]. Acad. Sci. USA 72, 2284-2288. Kondo, T., Fitzgerald, D.J., Chaudhary, V.K., Adhya, S. and Pastan, I. (1988) Activity of immunotoxins constructed with modified Pseudomonas exotoxin A lacking the cell recognition domain. J. Biol. Chem. 263(19), 9470-9475. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680-685. Leppla, S.H. (1976) Large-scale purification and characterization of the exotoxin of Pseudomonas aeruginosa. Infect. Immun. 14, 1077-1086. Lory, S. and Collier, R.J. (1980) Expression of enzymic activity by exotoxin A from Pseudomonas aeruginosa. Infect. Immun. 28, 494-501. Pastan, I., Chaudhary, V. and Fitzgerald, D.J. (1992) Recombinant toxins as novel therapeutic agents. Annu. Rev. Biochem. 61, 331-354. Ryan, L.D. and Vestling, C.S. (1974) Rapid purification of lactate dehydrogenase from rat liver and hepatoma: a new approach. Arch. Biochem. Biophys. 160, 279-284. Scopes, R.K. (1986) Strategies for enzyme isolation using dye-ligand and related adsorbents. J. Chromatogr. 376, 131-140.