Intranasal administration of an Escherichia coli-expressed codon-optimized rotavirus VP6 protein induces protection in mice

Protein Expression and PuriWcation 38 (2004) 205–216 www.elsevier.com/locate/yprep

Intranasal administration of an Escherichia coli-expressed codon-optimized rotavirus VP6 protein induces protection in mice Anthony H.-C. Choia,b,¤, Mitali Basua, Monica M. McNeala, Judy A. Beanc, John D. Clementsd, Richard L. Warda a

Division of Infectious Diseases, Cincinnati Children’s Hospital Medical Center, 3333 Burnet Avenue, Cincinnati, OH 45229-3039, USA b Phase 2 Discovery, Inc., 3130 Highland Avenue, Cincinnati, OH 45219-2374, USA c Department of Biostatistics, Cincinnati Children’s Hospital Medical Center, 3333 Burnet Avenue, Cincinnati, OH 45229-3039, USA d Department of Clinical Immunology, Tulane University Medical Center, New Orleans, LA 70112-2699, USA Received 1 June 2004, and in revised form 6 August 2004 Available online 12 October 2004

Abstract We are developing rotavirus vaccines based on the VP6 protein of the human G1P [J. Virol. 73 (1999) 7574] CJN strain of rotavirus. One prototype candidate consisting of MBP::VP6::His6, a chimeric protein of maltose-binding protein, VP6 and hexahistidine, was expressed mainly as truncated polypeptides in Escherichia coli BL21(DE3) cells. A possible reason for this extensive truncation is the high frequencies of rare bacterial codons within the rotavirus VP6 gene. Expression of truncated recombinant VP6 was found to be reduced, and expression of complete VP6 protein was simultaneously increased, when the protein was expressed in Rosetta(DE3)pLacI E. coli cells that contain increased amounts of transfer RNAs for a selection of rare codons. The same observation was made when a synthetic codon-optimized CJN-VP6 gene was expressed in E. coli BL21 or Rosetta cells. To increase protein recovery, recombinant E. coli cells were treated with 8 M urea. Denatured, full-length MBP::VP6::His6 protein was then puriWed and used for intranasal vaccination of BALB/c mice (2 doses administered with E. coli heat-labile toxin LT(R192G) as adjuvant). Following oral challenge with the G3P [J. Virol. 76 (2002) 560] EDIM strain of murine rotavirus, protection levels against fecal rotavirus shedding were comparable (P > 0.05) between groups of mice immunized with denatured codon-optimized or native (not codon-optimized) immunogen with values ranging from 87 to 99%. These protection levels were also comparable to those found after immunization with non-denatured CJN VP6. Thus, expression of complete rotavirus VP6 protein was greatly enhanced by codon optimization, and the protection elicited was not aVected by denaturation of recombinant VP6.  2004 Elsevier Inc. All rights reserved. Keywords: Rotavirus vaccine; Synthetic codon-optimized gene; Rare codons; Recombinant protein; Truncated polypeptides; Rosetta(DE3)pLacI E. coli cells; Urea denaturation; UltraWltration; AYnity chromatography; Enhanced expression; Retension of vaccine eYcacy

Rotavirus infections are the primary cause of severe gastroenteritis in young children and are responsible for ca. 500,000 deaths worldwide each year [1]. Use of eVective vaccines appears to be the only reliable method to reduce rotavirus-related mortality. In 1998, the Wrst and only FDA approved rotavirus vaccine, Rotashield, was

*

Corresponding author. Fax: +1 513 636 5466. E-mail address: [email protected] (A.H.-C. Choi).

1046-5928/$ - see front matter  2004 Elsevier Inc. All rights reserved. doi:10.1016/j.pep.2004.08.009

recommended for routine childhood immunization in the United States but was withdrawn in less than one year after being associated with intussusception, a rare form of bowel blockage [2]. Several other live, attenuated rotavirus vaccine candidates have been evaluated in clinical trials [3], but there is still a high level of concern about their eYcacies and possible associations with intussusception or other adverse events [4–6]. To alleviate safety concerns speciWc for live rotavirus vaccines, we are developing second generation vaccines

206

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

composed of Escherichia coli-expressed recombinant rotavirus VP6 proteins and adjuvant. VP6 is the sole structural protein of the middle of three capsid layers of rotavirus. It contains 397 amino acids and has a theoretical molecular mass of 45 kDa. It is the most conserved, immunogenic, and abundant rotavirus protein [7]. We have evaluated an experimental vaccine formulated with the recombinant MBP::VP6 protein, a chimera composed of the murine rotavirus strain EDIM VP6 genetically fused to the E. coli maltose-binding protein, and the potent mucosal adjuvant LT(R192G), a mutated form of E. coli heat-labile enterotoxin [8–12], whose toxicity is greatly attenuated [13]. The eYcacy of this formulation has been extensively evaluated in an adult mouse model in which vaccine eYcacies are measured as reductions in fecal rotavirus shedding in immunized mice relative to shedding in unimmunized mice following an oral EDIM challenge [8–12]. The formulation is extremely eYcacious, inducing between 95 and >99% reductions in rotavirus shedding in several inbred and outbred strains of mice, when delivered intranasally (i.n.) or orally [11]. One i.n. immunization of this formulation is suYcient to induce the maximum level of protection and this level of protection is maintained for at least one year [11]. In addition, a number of protective epitopes have been mapped in the EDIM VP6 protein using synthetic peptides to immunize mice belonging to H-2b and H-2d haplotypes [9,12]. Remarkably, a 14-amino-acid peptide containing a putative H-2d-restricted CD4 T cell epitope was found to induce a protection level comparable to that of puriWed recombinant VP6 proteins in H-2d BALB/c mice [9,12]. We are developing a vaccine composed of an E. coliexpressed chimeric VP6 protein derived from the human CJN strain of rotavirus for eventual use in humans. The CJN and EDIM VP6 proteins shared 91% identity in their amino acid sequence, a diVerence greater than those found among human rotavirus strains. The ability of VP6 proteins derived from the human G1P[8] CJN strain of rotavirus to protect against challenge by the heterotypic G3P[16] EDIM strain of murine rotavirus has been investigated. When administered intranasally along with LT(R192G), MBP::CJN-VP6 was found to generate protective eYcacies equivalent to those induced by MBP::EDIM-VP6 against an oral EDIM challenge, indicating that if this vaccine candidate is protective in humans, a single human rotavirus VP6 protein may elicit protection against all human rotavirus strains belonging to multiple serotypes [11]. The recombinant CJN and EDIM VP6 proteins used in our studies have all been puriWed using an maltosebinding protein (MBP) binding resin under non-denaturing conditions and contain small amounts of full-length MBP::VP6 but large quantities of truncated MBP::VP6 polypeptides [8]. In the studies reported herein, codon usage analysis was carried out to determine whether the

poor quality of expressed recombinant VP6 protein was the result of incompatible codon usage between VP6 and E. coli genes. Expression of full-length chimeric VP6 proteins was greatly increased by overcoming problems associated with incompatible codon usage in E. coli cells. These bacterial cells were also subjected to protein puriWcation conditions that improved the recovery of their expressed chimeric VP6 proteins but also caused protein denaturation. Finally, the protective eYcacy of puriWed chimeric VP6 proteins was determined after i.n. administration to mice together with the powerful mucosal adjuvant LT(R192G).

Materials and methods Synthesis of a codon-optimized VP6 gene to reduce usage of rare codons in E. coli cells Examination of codon usage in E. coli revealed greater usage of rare codons in the CJN VP6 gene than usage in E. coli genes (Table 1). A computer program available from Entelechon GmbH (Regensburg, Germany) and an E. coli codon usage table provided by Dr. Yasukazu Nakamura (Kazuka DNA Research Institute, Chiba, Japan) were used for backtranslation of the CJN VP6 protein sequence into a gene sequence in which usage of rare codons was reduced to mimic usage by E. coli genes (Fig. 4). Two sets of oligonucleotides (42-mers or smaller), one containing 28 contiguous oligonucleotides encompassing the sense strand and the other containing 28 contiguous oligonucleotides encompassing the antisense strand of the optimized CJN-VP6 sequence, were designed basing on the method described by Withers-Martinez [14] and were synthesized by the University of Cincinnati DNA core facility (Cincinnati, OH). In addition, six histidine codons were introduced in front of the stop codon in the oligonucleotide that corresponds to the 3⬘-end of the codon-optimized VP6 gene. The two sets of oligonucleotides were assembled into double-stranded polynucleotides and ampliWed by PCR to generate the synthetic codon-optimized CJN VP6::His6 gene. Construction of plasmids encoding native and codon-optimized VP6 The E. coli-expression plasmid pMAL-c2X (New England Biolabs, Beverly, MA) was used for expressing the chimeric VP6 proteins used in this study. This plasmid encodes the IPTG-inducible MBP::LacZ
⬘, a chimeric protein containing maltose-binding protein (MBP) and the LacZ
⬘ polypeptide. To construct the expression plasmid pMAL-c2X/CJNVP6-His6 that encodes the native CJN VP6-His6 sequence and expresses MBP::VP6::His6, PCR was carried out to generate

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

207

Table 1 Codon usage analysis of the E. coli, native VP6, codon-optimized VP6 genes Codons

E. coli

Native VP6

Frequencya

Frequencyb

Optimized VP6 No. of codons

Frequencyb

No. of codons

Ala GCA GCU GCG GCC

21 16 36 27

35 45 17 3

10 13 5 1

21 17 34 28

6 5 10 8

Cys UGU UGC

44 56

67 33

2 1

33 67

1 2

Asp GAU GAC

63 37

82 18

14 3

65 35

11 6

Glu GAA GAG

69 31

75 25

15 5

65 35

13 7

Phe UUU UUC

57 43

73 27

19 7

58 42

15 11

Gly GGA GGU GGG GGC

11 34 15 40

53 32 0 15

10 6 0 3

11 37 16 37

2 7 3 7

His CAU CAC

57 43

100 0

5 0

60 40

3 2

Ile AUA AUU AUC

7 51 42

28 72 0

8 21 0

10 48 41

3 14 12

Lys AAA AAG

77 23

50 50

4 4

75 25

6 2

Leu UUA UUG CUA CUU CUG CUC

13 13 4 10 50 10

45 17 9 17 9 3

16 6 3 6 3 1

14 11 6 11 49 9

5 4 2 4 17 3

Met AUG

100

100

12

100

12

Asn AAU AAC

45 55

92 8

33 3

47 53

17 19

Pro CCA CCU CCG CCC

19 16 53 12

70 20 10 0

14 4 2 0

20 15 50 15

4 3 10 3

Gln CAA CAG

35 65

76 24

16 5

33 67

7 14

Arg AGA

4

56

14

4

1 (Continued on next page)

208

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

Table 1 (continued) Codons AGG CGA CGU CGG CGC

E. coli Frequency 2 6 38 10 40

Native VP6 a

Frequency 16 0 20 4 4

b

Optimized VP6 No. of codons 4 0 5 1 1

Frequencyb 4 8 36 12 36

No. of codons 1 2 9 3 9

Ser AGU AGC UCA UCU UCG UCC

15 28 12 15 15 15

4 9 52 35 0 0

1 2 12 8 0 0

17 26 13 17 13 13

4 6 3 4 3 3

Thr ACA ACU ACG ACC

13 17 27 43

34 55 3 7

10 16 1 2

14 17 28 41

4 5 8 12

Val GUA GUU GUG GUC

15 26 37 22

32 48 8 12

8 12 2 3

16 28 36 20

4 7 9 5

Trp UGG

100

100

5

100

5

Tyr UAU UAC

57 43

70 30

7 3

60 40

6 4

100

100

1

100

1

Stop a

The frequencies of the codons of each amino acid used by 10,136 coding sequences. The frequencies of the codons of each amino acid present in the native and codon-optimized VP6 gene sequences are compared with those of 10,136 E. coli coding sequences. b

amplicons containing the native VP6 gene sequence (GenBank Accession No. AF461757) using the plasmid pMAL-c2X/CJNVP6 as the template, a forward primer (ATG GAG GTT TTA TAC TCA TTG), and a reverse primer containing six histidine codons (underlined) (TCA GTG ATG GTG ATG GTG ATG CTT AAT CAA CAT GCT TC), Platinum Pfx DNA Polymerase (Invitrogen/Life Technologies, Carlsbad, CA), and dNTP [11]. The synthetic DNAs containing either the native or codon-optimized VP6 coding sequence were cloned by blunt-end ligation into the XmnI site of pMAL/c2X, which follows the E. coli MBP-encoding sequence but precede the LacZ
⬘ gene sequence, to create the plasmids pMAL-c2X/CJNVP6-His6 and pMALc2X/CJNOptVP6-His6.

were used for expressing codon-optimized VP6 proteins. Strain BL21(DE3) is deWcient in lon and ompT proteases. Rosetta(DE3)pLacI was derived from Tuner(BL21 lacZY) that harbors the pRARE plasmid that encodes the tRNA genes of the rare codons for Arg (AGG, AGA), Gly (GGA), Ile (AUA), Leu (CUA), and pro (CCC). The recombinant plasmids pMAL-c2X, pMALc2X/CJNVP6-His6, and pMAL-c2X/CJNOptVP6-His6 were transformed into E. coli cells using standard methods [15]. Following transformation, cells were grown on nutrient agar plates containing antibiotics (100 g carbenicillin per liter for BL21(DE3), and 34 g chloramphenicol for Rosetta(DE3)pLacI cells). Single antibiotic-resistant recombinant colonies were selected for growth kinetic studies and protein expression.

Escherichia coli cell strains used for expressing recombinant VP6

Growth kinetics of E. coli BL21(DE3) cells expressing recombinant MBP::VP6::His6 or MBP::LacZ
⬘ protein

The E. coli cell strain BL21(DE3) (Stratagene, La Jolla, CA) was used to express native recombinant VP6, and Rosetta(DE3)pLacI cells (Novagen, Madison, WI)

The growth characteristics of recombinant E. coli cells containing the plasmids pMAL-cX2, pMAL-c2X/ CJNVP6-His6, or pMAL-c2X/CJNOptVP6-His6 that

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

express MBP::LacZ
⬘, MBP::VP6::His6 or MBP::OptVP6::His6, respectively, were determined. A single recombinant E. coli colony was used to inoculate a starter culture (5 mL of rich medium containing 10 g Bacto-tryptone, 5 g Bacto-yeast extract, 10 g NaCl, pH 7.0, and antibiotics appropriate for each bacteria strain). The cultures were grown in a shaker-incubator (37 °C, 2500 rpm) until the cell density reached 1 OD600 nm and were subsequently used to inoculate 50 mL of rich medium (starting OD600 nm of 0.05). When the OD600 nm reached approximately 0.5, IPTG (Wnal concentration of 1 mM) was added to induce expression of chimeric VP6 proteins. Aliquots of cells were withdrawn at 30-min intervals for determining increases in cell density (spectrophotmetric measurement at OD600 nm) and cell viability (plating on nutrient agar plates) during a 6-h period. Cell samples were serially diluted for optical density measurements (2-fold dilutions) and plating on agar plates (10-fold dilutions). The numbers of colonies on agar plates were counted after 24 h of incubation at 37 °C. The actual optical densities and viable cell concentrations were computed using the dilutions employed. PuriWcation of complete recombinant VP6 proteins Three hours after IPTG induction as described above, cells were harvested by centrifugation (4000g, 20 min, 4 °C) and frozen (¡80 °C) as cell pellets. The frozen pellets were thawed, resuspended in 100 mL of 8 M urea in PBS (50 mM sodium phosphate, and 300 mM NaCl, pH 7) and homogenized by repetitive passage through a 50mL syringe. The homogenates were then heated in a water bath (95 °C, 10 min) to further promote protein solubilization, then centrifuged (10,000g, 4 °C, 30 min) to remove insoluble materials. The supernatants were added to 10-mL volumes of packed polyhistidine-binding Talon resin (Clontech, Palo Alto, CA), which had been washed in PBS–urea. The chimeric VP6 proteins were allowed to bind to the resin on an orbital platform shaker (25 °C, 20 h). The resin was pelleted by centrifugation (700g, 5 min) and the supernatant was removed. The resin was then resuspended in 10 bed-volumes of PBS– urea and shaken on an orbital platform shaker. The bound VP6 protein was eluted by adding Wve bed-volumes of PBS–urea containing EDTA (100 mM, pH 8) and rocking on an orbital platform shaker (25 °C, overnight), and then separated from the resin by centrifugation (700g, 5 min). The elution step was repeated once by rocking the resin in Wve bed-volumes of PBS–urea– EDTA on an orbital platform rocker for 30 min. The eluates containing VP6 protein were pooled, concentrated, and buVer-exchanged using 1.5 L of PBS with a Stirred-cell Concentrator (Millipore, Bedford, MA). The yield of VP6 was determined using the Bradford assay (Amresco, Solon, OH ) and analyzed by Western blot analysis.

209

Western blot analysis of chimeric rotavirus proteins The quality of chimeric VP6 proteins expressed in E. coli cells and in Talon-resin-puriWed samples was analyzed by SDS–PAGE. Cells were collected 3 h after IPTG induction and cell densities were determined by spectrophotometry (OD600 nm). Cell samples were diluted to a concentration equivalent to 0.02 OD600 nm in gel loading buVer (50 mM Tris, pH 6.8; 5% SDS, 5%  mercaptoethanol, and 0.005% bromophenol blue, containing 10% w/v glycerol or 8 M urea), heated to 95 °C for 10 min, cooled to room temperature, and then subjected to SDS–PAGE. The separated proteins in the SDS–gels were blotted onto nitrocellulose sheets, which were blocked with 5% skim milk in TBS (25 mM Tris– HCl, and 0.9% NaCl, pH 7.5). The sheets were incubated with a rabbit anti-MBP (New England Biolabs), a rabbit polyclonal anti-human group A rotavirus (DAKO, Carpentaria, CA), or rabbit anti-polyhistidine antibody (Santa Cruz Biotechnology, Santa Cruz, CA). After washing with TTBS (0.1% Tween 20 in TBS), the sheet was incubated with goat anti-rabbit IgG conjugated to alkaline phosphatase (Invitrogen/Life Technologies). The sheets were washed with TTBS and then incubated with 5-bromo-4-chloro-3-indolylphosphate (BCIP, Invitrogen/Life Technologies) and nitroblue tetrazolium (NBT, Invitrogen/Life Technologies) to visualize antibody-bound polypeptides. Quantitative Western blot analysis was carried out to estimate the amounts of expressed VP6 protein in E. coli cells collected 3 h post-IPTG induction. Cell samples and puriWed MBP (100 ng in 10 L, New England BioLabs) were serially (2-fold) diluted in SDS–gel sample buVer containing 8 M urea and subjected to Western blot analysis. The lowest concentrations at which MBP-containing VP6 and MBP were still detectable in the blots were noted and these concentrations were used for estimating the amounts of expressed VP6 in the VP6-containing samples. Mice Female BALB/c mice were purchased from HarlanSprague–Dawley (Indianapolis, IN) and used at 6–8 weeks of age. None of the animals had evidence of rotavirus antibody as determined by ELISA. Microisolation cages were used to house the mice and all procedures were conducted in accordance with protocols approved by Cincinnati Children’s Hospital Research Foundation Institutional Animal Care and Use Committee. Immunization Mice were anesthetized in a closed vessel with isoXurane (Abbott Laboratories, Chicago, IL). Groups of mice (8 mice per group) were i.n. immunized (20 L per dose,

210

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

10 L per nostril; two immunizations separated by 2 weeks) with 10 g of chimeric VP6 expressed from either the native or the codon-optimized VP6 gene in BL21(DE3) and Rosetta cells(DE3)pLacI, respectively, that were admixed with 10 g of LT(R192G) as adjuvant. A group of unimmunized mice served as the control. Challenge of mice with EDIM rotavirus Four weeks after the last immunization, mice were orally (gavage) challenged with 1000 Shedding-Dose-50 of unpassaged EDIM rotavirus. Detection of rotavirus antigen in stools Two to three fecal pellets were collected daily from each mouse into Earle’s balances salt solution (EBSS) starting on day 1 following rotavirus challenge and for the next 6 days. Samples were stored frozen as 10% (w/v) suspensions in EBSS until analyzed, at which time they were homogenized and centrifuged (1500g, 5 min, 4 °C) to remove debris. Quantities of rotavirus antigen in the fecal samples were determined by ELISA as ng/mL of homogenate, using methods already described [8–12,16]. Statistical methods The Bonferroni t test (SAS Software, SAS Institute, NC) was used to compare the quantities of rotavirus shed in stool samples among the mouse groups. The sum of the rotavirus antigen shed in stool samples for each mouse over the 7-day post-immunization period were used for computation. DiVerences between samples were considered signiWcant when the probability levels (P) were <0.05.

ing MBP::VP6::His6 containing six histidine amino acid residues was constructed. Cell samples expressing the two MBP-containing proteins were collected 3 h postIPTG induction, and triplicates of the cell samples were subjected to SDS–PAGE. The separated proteins were then transferred onto nitrocellulose paper. To visualize complete and truncated chimeric VP6 polypeptides, the triplicate blots were probed with rabbit polyclonal antiHis6, anti-human group A rotavirus, or anti-MBP antibody (Figs. 1A–C, respectively). The blots were then probed with goat anti-rabbit antibody conjugated to alkaline phosphatase, and then with the substrates BCIP/NBT. Two protein bands that probably represent two conformations of recombinant VP6 were recognized by all three primary antibodies (Figs. 1A–C; lane 2). In the blots probed with anti-MBP antibodies, additional multiple truncated VP6 polypeptides were detected (Fig. 1C, lane 2), whereas MBP::LacZ
⬘ was expressed almost solely as a single complete protein (Fig. 1C, lane 1). When probed with anti-human group A rotavirus antibody, multiple truncated VP6 proteins were also detected (Fig. 1B, lane 2) but, when probed with antiHis6 antibodies, no truncated polypeptides were detected in the blot containing recombinant VP6 proteins (Fig. 1A, lane 2). These results suggested that the truncated MBP::VP6::His6 proteins contain the N-terminal MBP but lack various portions of the C-terminus. Thus, these results show that expression of truncated MBP-containing polypeptides was inherent with expression of VP6, and MBP was not responsible for truncation of the MBP::VP6::His6 protein.

Results MBP::VP6::His6 is primarily expressed as a truncated recombinant protein We recently evaluated in an adult mouse model a rotavirus vaccine candidate composed of a recombinant E. coli-expressed VP6 protein that was derived from the human rotavirus strain CJN [11]. The recombinant VP6 protein was encoded by the plasmid pMAL-c2X and expressed in BL21(DE3) E. coli cells as MBP::VP6, a chimeric protein that contains the plasmid-encoded maltose-binding protein at its amino terminus linked to the complete CJN VP6. Because VP6 was expressed mainly as truncated proteins [8–11], it was of interest to determine whether MBP::LacZ
⬘, a chimeric protein encoded by the parental plasmid is also expressed as truncated proteins. In order to detect as well as purify complete VP6 proteins, a recombinant plasmid express-

Fig. 1. Western blot analysis of recombinant VP6 derived from the human CJN rotavirus strain. Recombinant BL21(DE3) E. coli cells containing the plasmid pMAL-c2X/CJNVP6-His6, which expresses the MBP::CJNVP6::His6 protein, or the parental plasmid pMAL-c2X expressing MBP::LacZ
⬘ were grown in the presence of 1 mM IPTG for 3 h. Cells (3 mL) were harvested and resuspended in gel loading buVer containing glycerol (200 L). Triplicate sets of cell samples (20 L) were subjected to Western blot analysis. The blots were probed with either the primary rabbit anti-His6 (A), anti-human rotavirus (B), or anti-MBP (C) antibodies, followed by the secondary antibody goat anti-rabbit conjugated to alkaline phosphatase and Wnally the substrates BCIP/NBT. Lane 1: cell lysate containing MBP::LacZ
⬘ protein. Lane 2: cell lysate containing MBP::CJNVP6::His6 protein. The size of the molecular weight markers (kDa) are indicated to the left of the blots. The open arrow denotes complete MBP::LacZ
⬘ and the solid arrows denote complete MBP::CJNVP6::His6.

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

Overexpression of MBP::VP6::His6 or MBP::LacZ
⬘ reduces cell viability Because complete MBP::VP6 protein was expressed in lower amounts than MBP::LacZ
⬘ (Fig. 1), we investigated whether expressed recombinant VP6 was toxic to E. coli cells. The growth characteristics of E. coli cells expressing these recombinant proteins were determined, either in the absence or presence of IPTG during a 6-h growth period, using spectrophotometric measurements (OD600 nm) of shaken cultures and by determining colony counts on nutrient agar plates. The former method measures both viable and non-viable cells while the latter measures only viable cells (Fig. 2). Whether incubated in the absence or presence of IPTG, the cell densities of the two shaken cultures increased with similar kinetics. The viable cell counts of both cultures also increased steadily in the absence of IPTG. In contrast, the increase in viable cell counts occurred only during the initial 3 h in the presence of IPTG, after which time the viable cell numbers decreased in all three cultures with similar kinetics. Thus, overexpression of MBP::VP6::His6 or MBP::LacZ
⬘ had similar eVects on cell viability. Such eVects have been linked to the depletion of tRNAs during protein expression in E. coli [17]. From these results, we conclude that expression of MBP::VP6::His6 did not have a diVerential toxic eVect on E. coli cells. The presence of truncated recombinant VP6 appears to be caused by codon usage problems The presence of truncated polypeptides in E. coli has been reported during overexpression of recombinant proteins when the genes encoding these proteins have much

211

higher frequencies for codons of arginine, isoleucine, leucine, glycine, and proline than found in E. coli coding sequences [18,19]. The frequencies of these codons in the CJN VP6 gene were computed, and were found to be much higher than the average found in >10,000 E. coli genes (Table 1). Perhaps not unexpectedly, usage of these codons by the E. coli LacZ
⬘ polypeptide was comparable to those of the E. coli coding sequences. For example, codon frequencies of the arginine AGG and AGA, isoleucine codons AUU and AUC, and glycine GGA codons (7, 7, 46, 54, and 11%, respectively) were more similar to those found in E. coli (2, 3, 51, 42, and 11%, respectively) than in the CJN VP6 gene (16, 56, 72, 0, and 52%, respectively). Thus, occurrence of truncated VP6 protein in E. coli was likely caused by codon bias in the CJN VP6 gene. Complete recombinant VP6 protein is expressed in greater quantities in Rosetta(DE3)pLacI cells than in BL21(DE3) cells Strains of E. coli that express tRNA genes of rare codons have been reported to overcome problems associated with codon bias [18,21,22]. Rosetta(DE3)pLacI, one such strain, contains the plasmid pRARE that encodes the tRNAs for the arginine AGG, AGA, isoleucine AUA, leucine CUA, glycine GGA, and proline CCC codons. The codons recognized by four out of these six tRNAs are rare in E. coli gene sequences but are prominent within the CJN VP6 protein gene. To determine whether the quality of MBP::VP6::His6 could be improved by expression in Rosetta(DE3)pLacI cells, both recombinant Rosetta(DE3)pLacI and BL21(DE3) cells were grown in the presence of IPTG. After 3 h, cells were collected by centrifugation and solubilized in SDS–

Fig. 2. Growth kinetics of E. coli cells overexpressing recombinant MBP-containing VP6 or LacZ
⬘ proteins. Recombinant BL21(DE3) E. coli cells expressing MBP::CJNVP6::His6 encoded by the native VP6 gene (B and E), MBP::CJNOptVP6::His6 encoded by the codon-optimized VP6 gene (C and F), or MBP::LacZ
⬘ encoded by the parental plasmid used for expressing recombinant VP6 proteins (A and D), were grown in the presence of IPTG (1 mM) for 6 h. Cells were sampled every 1.5 h, and serially diluted (2-fold) for optical density (OD600 nm ) determination (upper panel). Simultaneously, the cell samples were 10-fold serially diluted and plated in triplicates on nutrient agar plates, which were incubated at 37 °C for 24 h. The number of colonies were counted and the concentrations of viable cells were calculated (lower panel). The open and closed circles represent growth in the absence or presence of IPTG.

212

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

gel sample buVer containing either glycerol or urea. Duplicate sets of cell samples were subjected to SDS– PAGE, and separated proteins were transferred onto nitrocellulose sheets. The blots were probed with antiMBP or anti-His6 antibodies to detect complete and truncated recombinant VP6 proteins. The quantities of complete VP6 expressed in Rosetta(DE3)pLacI cells were noticeably greater than those in BL21(DE3) cells (Fig. 3). Concomitantly, smaller quantities of truncated polypeptides were observed in Rosetta(DE3)pLacI cells than in BL21 cells. The amounts of protein detected were higher when urea was used rather than glycerol in SDS– gel sample buVer, suggesting that the VP6 protein might be secluded within inclusion bodies that are typically present when recombinant proteins are overexpressed [23]. When quantitative Western blot analyses of 2-fold serially diluted cell samples and MBP standard were carried out, the amounts of complete MBP::VP6::His6 expressed in Rosetta(DE3)pLacI cells was estimated to be between 10 and 20 mg/L of cells (results not shown). This was 5- to 10-fold greater than the amount expressed in BL21(DE3) cells. Therefore, the amount of complete chimeric MBP::VP6::His6 was increased and the amount of truncated MBP-containing polypeptides was concomitantly reduced by increasing the availability of rare tRNA codons. Quantities of complete recombinant VP6 protein expressed from a codon-optimized VP6 sequence is higher in either BL21(DE3) or Rosetta(DE3)pLacI E. coli cells than expressed from the native VP6 sequence in BL21(DE3) cells Truncated MBP-containing VP6 was still expressed in Rosetta(DE3)pLacI cells (Figs. 3A and B, lane 2), albeit in smaller quantities than that in BL21(DE3) (Figs. 3A and B, lane 1). Inspection of the native VP6 gene sequence revealed that rare E. coli codons, in addition to

those that are well-known, are present in high frequencies (Table 1). Notably, alanine GCA and GCU, histidine CAU, lysine AAG, asparigine AAU, glutamine CAA, serine UCA, UCU, threonine ACA, ACU, and valine GUA, GUU account for considerably higher usage in CJN VP6 (34, 45, 100, 50, 92, 76, 52, 35, 34, 55, 32, and 48%, respectively) than in E. coli (21, 16, 0, 23, 8, 35, 12, 15, 13, 17, 15, and 26%, respectively). As expected, when the MBP::LacZ
⬘ gene was analyzed, codon usage was similar to that used by other E. coli coding sequences (results not shown). Therefore, we determined whether a VP6 gene sequence, which consists of codons whose frequencies mimic those found in E. coli coding sequences, would reduce VP6 truncation in BL21(DE3) cells and further reduce truncation in Rosetta(DE3)pLacI cells. To do this, the native VP6 protein sequence was backtranslated, using an E. coli codon usage table, into a sequence (Fig. 4) whose codons reXect usage in the E. coli genome (Table 1). Using this codon-optimized sequence, contiguous synthetic oligonucleotides encompassing both plus and negative strands of the entire gene were synthesized. The oligonucleotide encoding the 3⬘-end of the gene was generated to contain six histidine codons immediately before the stop codon. Using PCR, these oligonucleotides were used to assemble the complete VP6 protein gene that was then cloned into the plasmid pMAL-c2X creating pMAL-c2X/OptVP6-His6. The plasmid was transformed into Rosetta(DE3)pLacI, as well as BL21(DE3) cells, which were grown in the presence of IPTG for 3 h. Rosetta(DE3)pLacI or BL21(DE3) cells containing recombinant VP6 expressed from either the synthetic or native gene were subjected to SDS–PAGE. The separated proteins were transferred to duplicate blots, which were probed with anti-MBP (Figs. 3A and C) or anti-His6 antibodies (Figs. 3B and D). The quantity of full-length chimeric VP6 expressed from the codon-optimized VP6 gene was greater in Rosetta(DE3)pLacI cells than VP6 expressed from the native gene in BL21(DE3) cell, but

Fig. 3. Expression of recombinant VP6 encoded by the native or codon-optimized CJN VP6 coding sequences in the BL21(DE3) and Rosetta(DE3)pLacI strains of E. coli. Recombinant BL21(DE3) and Rosetta(DE3)pLacI cells were grown in shaken cultures in the presence of IPTG (1 mM) for 3 h at 37 °C. Aliquots of cells were withdrawn and optical densities (OD600 nm) were measured. The cells were diluted to a Wnal concentration of 0.02 OD600 nm. Ten microliters of the diluted samples were subjected to SDS–PAGE and then Western blot analysis. Panels A and B: glycerol (10%) was used in the SDS–PAGE sample buVer. Panels C and D: urea (8 M) was used in the SDS–PAGE sample buVer. Lane 1: native VP6 expressed in BL(21) cells; lane 2: native VP6 expressed in Rosetta(DE3)pLacI cells; lane 3: codon-optimized VP6 expressed in BL21(DE3)pLacI cells; lane 4: codon-optimized VP6 expressed in Rosetta(DE3)pLacI cells. Panels A and C: rabbit anti-MBP antibodies were used as the primary antibody; Panels B and D: rabbit anti-polyhistidine antibodies were used as the primary antibody. The arrows indicate full-length MBP::VP6 proteins.

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

213

Fig. 4. Codon-optimized nucleotide sequence of CJN VP6 protein. The codon-optimized gene sequence (bottom sequence) was generated by backtranslation of the native CJN VP6 amino acid sequence (Accession No. AF461757) using an E. coli codon usage table and is shown with the native unmodiWed sequence (upper sequence). Only the nucleotides that were altered in the codon-optimized gene sequence are shown.

the amount of full-length VP6 expressed from the optimized gene in Rosetta(DE3)pLacI cells did not appear to be diVerent from that expressed in BL21(DE3) cells. PuriWcation of urea and heat-denatured chimeric codonoptimized VP6 using one-step aYnity resin puriWcation A greater amount of complete recombinant codonoptimized VP6 protein could be detected in Western blots when cells were solubilized in 8 M urea and heated for 10 min at 98 °C (Figs. 3C and D). Since no His6-containing truncated VP6 proteins were observed (Fig. 3D), it should be possible to purify complete recombinant VP6 using a resin that binds His6-containing proteins. Rosetta(DE3)pLacI cells expressing codon-optimized VP6 protein were harvested 3 h post-IPTG induction and were solubilized in PBS containing urea (8 M) at 98 °C for 10 min. Recombinant VP6 was puriWed from the cell lysate using the His6-binding Talon resin, urea was removed, and VP6 protein was concentrated by ultraWltration in PBS. Western blot analyses was performed, and only complete MBP::VP6::His protein was

detected using either anti-MBP or anti-His6 antiserum (results not shown). Complete recombinant VP6 puriWed in the presence of urea induced excellent protection in mice The protective eYcacies of complete chimeric VP6 proteins encoded by the codon-optimized VP6 gene and expressed in Rosetta(DE3)pLacI, and from the native VP6 gene expressed in BL21(DE3) were evaluated in BALB/c mice. The proteins were delivered i.n. (two immunizations of 10, 1, 0.1, or 0.01 g per dose separated by 2 weeks) together with LT(R192G) (10 g). Four weeks after the second immunization, mice were challenged with an oral dose of 1000 SD50 of EDIM. Both the 10 and 1 g doses of complete VP6 expressed from the codon-optimized gene induced excellent protection (99 and 96%, respectively; P < 0.05; Fig. 5). VP6 expressed from the native gene induced less but signiWcant (P < 0.05) protection in this study (87 and 91%, respectively) but this protection was not signiWcantly (P > 0.05) diVerent from that induced by VP6 expressed

214

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

Fig. 5. Protection from fecal shedding in mice immunized with recombinant VP6 encoded by the native or codon-optimized VP6 gene. Recombinant MBP::CJNVP6::His6 and MBP::CJNOptVP6::His6 protein were expressed from either the native or codon-optimized VP6 gene and puriWed in the presence of 8 M urea, which was subsequently removed by buVer exchange. BALB/c mice were administered intranasally (two immunizations, 2 weeks apart) either with 10, 1, 0.1, 0.01, or 0.001 g of chimeric VP6, followed by an oral challenge with EDIM rotavirus 4 weeks after the second immunization. Stool specimens were collected and assayed for the amount of rotavirus shed during the 7-day period after virus challenge. Quantities shed in the immunized groups were compared with those shed by the unimmunized control group and expressed as percent reduction in rotavirus shedding. The symbol (*) indicates that shedding was not signiWcantly (P > 0.05, Bonferroni’s ANOVA) diVerent among the groups compared. The symbol (+) denotes that shedding was signiWcantly diVerent between the two groups.

from the codon-optimized gene. However, immunization with 0.1 g of the codon-optimized VP6 did induce signiWcantly higher protection than the native VP6 protein (82 and 47%, respectively, P < 0.05). Further reductions in VP6 doses to 0.01 and 0.001 g resulted in no protection.

Discussion We have utilized our adult mouse rotavirus-challenge model to evaluate the protective eYcacies of recombinant rotavirus vaccines based on the VP6 protein, the protein that constitutes the intermediate capsid layer of the virus particle [8–12,16]. The VP6 protein used in these studies was derived either from the mouse G3P[16] EDIM (homologous) or human G1P[8] CJN (heterologous) rotavirus strains and expressed in E. coli BL21(DE3) cells. Intranasal or oral administration of recombinant EDIM or CJN VP6 proteins formulated with strong adjuvants induced between 86 and 99% protection against fecal rotavirus shedding in all strains of outbred or inbred mice examined following oral challenge with murine rotaviruses. Based on these encouraging results, we have continued our evaluations of VP6 as a vaccine candidate. In the development of E. coli-expressed VP6 as a vaccine candidate, it is important to maximize the yield of the full-length protein to be used for immunization. This is especially pertinent since E. coli-expressed chimeric EDIM and CJN VP6 proteins were shown to contain

large quantities of truncated VP6 polypeptides ([8], Fig. 1). It is unlikely that the truncations were caused by proteolytic degradation since the proteins were expressed in BL21(DE3), an E. coli strain that is deWcient in the Omp and Lon proteases. A similar pattern of VP6 truncation was also observed in SDS–PAGE gels when glutathione S-transferase was used as the fusion partner [24], thus suggesting that these truncations were associated with the VP6 protein itself rather than the fusion partners. Protein truncations have also been observed during overexpression of other recombinant proteins and have, at times, been attributed to codon-usage bias that occurs when heterologous proteins are expressed in host cells such as E. coli [25]. Codon bias arises because the use of the 61 mRNA codons varies between organisms. Generally, the frequency of codons in genome sequences reXects the abundance of cognate tRNAs [19]. For amino acids that utilize two or more synonymous codons, some codons called minor codons are rarely used [26]. In E. coli, the rare codons are the arginine codons AGG, AGA, CGA, CGG, the isoleucine codon AUA, the leucine codon CUA, the proline codon CCC, and the serine codon UCG [19]. When depletion of tRNAs that recognize minor codons occurs during overexpression of foreign proteins, growth rate decreases and synthesis of truncated polypeptides occurs [17,27]. The Wrst step toward enhancement of VP6 protein expression was to compare codon usage within the CJN VP6 protein gene with that of >10,000 coding sequences of E. coli. We found the codons AGG, AGA of arginine,

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216

AUA of isoleucine, and GGA of glycine were used much more frequently in the CJN VP6 gene than in the E. coli coding sequences (16, 56, 28, and 52 versus 2, 3, 7, and 11%, respectively). This supported the suggestion that poor expression of complete VP6 and high abundance of truncated VP6 polypeptides were due to codon bias. Expression of truncated VP6 protein was partially overcome in Rosetta(DE3)pLacI, an E. coli cell strain that supplies extra copies of E. coli tRNA genes encoding arginine AGG, AGA, glycine GGA, isoleucine AUA, leucine CUA, and proline CCC tRNAs. The codon bias problem was also alleviated when chimeric VP6 encoded by a synthetic CJN VP6 gene, whose codon usage reXects those of E. coli gene sequences, was expressed in recombinant BL21(DE3) E. coli cells. In fact, the quantities of full-length recombinant CJN VP6 expressed by the codon-optimized plasmid in either Rosetta(DE3)pLacI or BL21(DE3) cells were 5- to 10-fold greater than the quantities expressed from the native CJN VP6 gene in BL21(DE3) cells. Because truncations of VP6 were still observed after translation of the synthetic gene in Rosetta(DE3)pLacI cells, other measures are needed if this problem is to be completely eliminated. It has been reported that the socalled Class II proteins, unlike other E. coli proteins, are highly and continuously expressed even during exponential growth, and these proteins rarely use minor codons [28]. Therefore, production of a synthetic VP6 gene with the same codon usage as found in the genes for these proteins may eliminate truncation. In addition to codon usage, the nucleotide sequences that Xank the initiation codon and the presence of secondary mRNA structure have additional eVects on the quality and levels of protein expression [20,29–34]. Modifying the secondary structure of CJN VP6 mRNA or the nucleotides surrounding the initiation codon may further maximize full-length rotavirus VP6 expression in E. coli. We determined that >90% of expressed VP6 protein was sequestered in inclusion bodies within the E. coli cells (results not shown). Based on this, we were able to greatly increase the recovery of recombinant MPB::CJNVP6::His6 protein by solubilizing recombinant cells in urea (8 M) and using the His6-binding Talon aYnity resin for puriWcation. The vaccine eYcacies of chimeric CJN VP6 proteins, expressed from either the codon-optimized gene in Rosetta(DE3)pLacI E. coli cells or the native gene in BL21(DE3) cells and subsequently puriWed in the presence of 8 M urea, were compared in mice. The proteins were delivered by i.n. administration together with the adjuvant LT(R192G). Although protection elicited by the codon-optimized VP6 protein was somewhat greater at the higher dosage levels (10 or 1 g) than VP6 expressed from the native gene, the diVerence was not signiWcant until the dosages were decreased to 0.1 g. At this dosage, the codon-optimized VP6 was signiWcantly more eYcacious than the

215

native VP6 (82 versus 47%, P < 0.05). A possible explanation for the better protection by VP6 from the codonoptimized gene is that translation of VP6 from the native gene caused rapid depletion of the rare tRNAs. Therefore, during overexpression of the VP6 protein, frameshift, deletion, and point mutations occur more frequently [27,34], potentially altering the epitopes that are important for induction of protection. Based on the results reported here, production of large amounts of full-length chimeric VP6 protein in E. coli cells is a realistic goal that helps facilitate the development of VP6 as a vaccine candidate. Because the synthetic VP6 gene, still contains rare codons, albeit in lower frequencies, further modiWcations of the synthetic VP6 gene to completely remove all rare codons may increase the yield of this product.

Acknowledgment This work was supported in part by Public Health Service contract NO1AI 45252 from NIH-NIAID to Cincinnati Children’s Hospital Medical Center and Public Heath Service Grant R43AI50326 from NIH-NIAID to A.H. Choi.

References [1] U.D. Parashar, E.G. Hummelman, J.S. Bresee, M.A. Miller, R.I. Glass, Global illness and deaths caused by Rotavirus Disease in children, Emerg. Infect. Dis. 9 (2003) 565–572. [2] T.V. Murphy, P.M. Gargiullo, M.S. Massoudi, D.B. Nelson, A.O. Jumaan, C.A. Okoro, L.R. Zanardi, S. Setia, E. Fair, C.W. LeBaron, M. Wharton, J.R. Livingood, Rotavirus intussusception investigation team. Intussusception among infants given an oral rotavirus vaccine, N. Engl. J. Med. 344 (2001) 564–572. [3] R.L. Ward, Possible mechanisms of protection elicited by candidate rotavirus vaccines as determined with the adult mouse model, Viral Immunol. 16 (2003) 17–24. [4] P.N. Goldwater, K. Rowland, M.K. Thesinger, A. Abbott, E.A. Grieve, P.J. Palombo, I. Masendycz, I. Wilkinson, J. Bear, Rotavirus encephalopathy: pathogenesis reviewed, J. Paediatr. Child Health. 37 (2001) 206–209. [5] M.C. Honeyman, B.S. Coulson, N.L. Stone, S.A. Gellert, P.N. Goldwater, C.E. Steele, J.J. Couper, B.D. Tait, P.G. Colman, L.C. Harrison, Association between rotavirus infection and pancreatic islet autoimmunity in children at risk of developing type 1 diabetes, Diabetes 49 (2000) 1319–1324. [6] R.J. Sokol, C. Mack, Etiopathogenesis of biliary atresia, Semin. Liver Dis. 21 (2001) 517–524. [7] B. Tang, J.M. Gilbert, S.M. Matsui, H.B. Greenberg, Comparison of the rotavirus gene 6 from diVerent species by sequence analysis and localization of subgroup-speciWc epitopes using site-directed mutagenesis, Virology 237 (1997) 89–96. [8] A.H. Choi, M. Basu, M.M. McNeal, J.D. Clements, R.L. Ward, Antibody-independent protection against rotavirus infection of mice stimulated by intranasal immunization with chimeric VP4 or VP6 protein, J. Virol. 73 (1999) 7574–7581. [9] A.H. Choi, M. Basu, M.M. McNeal, J. Flint, J.L. VanCott, J.D. Clements, R.L. Ward, Functional mapping of protective domains

216

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17] [18]

[19] [20]

[21]

A.H.-C. Choi et al. / Protein Expression and PuriWcation 38 (2004) 205–216 and epitopes in the rotavirus VP6 protein, J. Virol. 74 (2000) 11574–11580. A.H. Choi, M.M. McNeal, J.A. Flint, M. Basu, N.Y. Lycke, J.D. Clements, J.D. Bean, H.L. Davis, M.J. McCluskie, J.L. VanCott, R.L. Ward, The level of protection against rotavirus shedding in mice following immunization with a chimeric VP6 protein is dependent on the route and the coadministered adjuvant, Vaccine 20 (2002) 1733–1740. A.H. Choi, M.M. McNeal, M. Basu, J.A. Flint, S.C. Stone, J.D. Clements, J.A. Bean, S.A. Poe, J.L. VanCott, R.L. Ward, Intranasal or oral immunization of inbred and outbred mice with murine or human rotavirus VP6 proteins protects against viral shedding after challenge with murine rotaviruses, Vaccine 20 (2000) 3310– 3321. A.H. Choi, M.M. McNeal, M. Basu, J.A. Bean, J.L. VanCott, J.D. Clements, R.L. Ward, Functional mapping of protective epitopes within the rotavirus VP6 protein in mice belonging to diVerent haplotypes, Vaccine 21 (2003) 761–767. B.L. Dickinson, J.D. Clements, Dissociation of Escherichia coli heat-labile enterotoxin adjuvanticity from ADP-ribosyltransferase activity, Infect. Immun. 63 (1995) 1617–1623. C. Withers-Martinez, E.P. Carpenter, F. Hackett, B. Ely, M. Sajid, M. Grainger, M.J. Blackman, PCR-based gene synthesis as an eYcient approach for expression of the A+T-rich malaria genome, Protein Eng. 12 (1999) 1113–1120. F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smit, K. Struhl (Eds.), Current Protocols in Molecular Biology, Greene Publishing Associates Inc., and John Wiley & Sons, Inc., New York, 1995. M.M. McNeal, J.L. VanCott, A.H. Choi, M. Basu, J.A. Flint S.C.Stone, J.D. Clements, R.L. Ward, CD4 T cells are the only lymphocytes needed to protect mice against rotavirus shedding after intranasal immunization with a chimeric VP6 protein and the adjuvant LT(R192G), J. Virol. 76 (2002) 560–568. V. Emilsson, C.G. Kurland, Growth rate dependence of transfer RNA abundance in Escherichia coli, EMBO J. 9 (1990) 4359–4366. A.M. Baca, W.G. Hol, Overcoming codon bias: a method for highlevel overexpression of Plasmodium and other AT-rich parasite genes in Escherichia coli, Int. J. Parasitol. 30 (2000) 113–118. S.P. Zhang, G. Zubay, E. Goldman, Low-usage codons in Escherichia coli, yeast, fruit Xy and primates, Gene 105 (1991) 61–72. D.B. Carlini, Y. Chen, W. Stephan, The relationship between third-codon position nucleotide content, codon bias, mRNA secondary structure and gene expression in the drosophilid alcohol dehydrogenase genes Adh and Adhr, Genetics 159 (2001) 623–633. M. Loyevsky, F. Mompoint, E. Yikilmaz, S.F. Altschul, T. Madden, J.C. Wootton, J. Kurantsin-Mills, O.O. Kassim, V.R. Gor-

[22]

[23]

[24]

[25]

[26] [27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

deuk, T.A. Rouault, Expression of a recombinant IRP-like Plasmodium falciparum protein that speciWcally binds putative plasmodial IREs, Mol. Biochem. Parasitol. 126 (2003) 231–238. M. Ubeidat, C.L. Rutherford, PuriWcation and renaturation of Dictyostelium recombinant alkaline phosphatase by continuous elution electrophoresis, Protein Expr. Purif. 27 (2003) 375–383. B.H. Rehm, Q. Qi, B.B. Beermann, H.J. Hinz, A. Steinbuchel, Matrix-assisted in vitro refolding of Pseudomonas aeruginosa class II polyhydroxyalkanoate synthase from inclusion bodies produced in recombinant Escherichia coli, Biochem. J. 358 (2001) 263–268. D.M. Banos, S. Lopez, C.F. Arias, F.R. Esquivel, IdentiWcation of a T-helper cell epitope on the rotavirus VP6 protein, J. Virol. 71 (1997) 419–426. M.J. Pikaart, G. Felsenfeld, Expression and codon usage optimization of the erythroid-speciWc transcription factor cGATA-1 in baculoviral and bacterial systems, Protein Expr. Purif. 8 (1996) 469–475. J.M. Comeron, M. Aguade, An evaluation of measures of synonymous codon usage bias, J. Mol. Evol. 47 (1998) 268–274. M.D. Forman, R.F. Stack, P.S. Masters, C.R. Hauer, S.M. Baxter, High level, context dependent misincorporation of lysine for arginine in Saccharomyces cerevisiae a1 homeodomain expressed in Escherichia coli, Protein Sci. 7 (1998) 500–503. A. Henaut, A. Danchin, in: F. Neidhardt, R. Curtiss III, J. Ingraham, E. Lin, B. Low, B. Magasanik, W. ReznikoV, M. Riley, M. Schaechter, H. Umbarger (Eds.), Escherichia coli and Salmonella typhimurium Cellular and Molecular Biology, vol. 2, American Society for Microbiology, Washington, DC, 1996, pp. 2047– 2066. I.L. Hofacker, M. Fekete, P.F. Stadler, Secondary structure prediction for aligned RNA sequences, J. Mol. Biol. 319 (2002) 1059– 1066. J.M. Fernandez, J.P. HoeZer (Eds.), Gene Expression Systems, Using Nature for the Art of Expression, Academic Press, San Diego, 1999. D.H. Mathews, D.H. Turner, Dynalign: an algorithm for Wnding the secondary structure common to two RNA sequences, J. Mol. Biol. 317 (2002) 191–203. M. Oresic, D. Shalloway, SpeciWc correlations between relative synonymous codon usage and protein secondary structure, J. Mol. Biol. 281 (1998) 31–48. C.M. Stenstrom, E. Holmgren, L.A. Isaksson, Cooperative eVects by the initiation codon and its Xanking regions on translation initiation, Gene 273 (2002) 259–265. K. Parker, Errors and alternative in reading the universal genetic code, Microbiol. Rev. 53 (1989) 273–398.