Soluble cytoplasmic expression, rapid purification, and characterization of cyanovirin-N as a His-SUMO fusion

Abstract

Cyanovirin-N (CVN) is a promising antiviral candidate that has an extremely low sequence homology with any other known proteins. The efficient and soluble expression of biologically functional recombinant CVN (rCVN) is still an obstacle due to insufficient yield, aggregation, and abnormal modification. Here, we describe an improved approach to preparing native rCVN from Escherichia coli more efficiently. A fusion gene consisting of cvn and sumo (small ubiquitin-related modifier) and a hexahistidine tag was constructed according to the codon bias of the host cell. This small ubiquitin-related modifier (SUMO)-fused CVN is expressed in the cytoplasm of E. coli in a folded and soluble form (>30% of the total soluble protein), yielding 3 to 4 mg of native rCVN from 1 g of wet cells to a purity up to 97.6%. Matrix-assisted laser desorption ionization coupled to time-of-flight mass spectrometry and reverse-phase high-performance liquid chromatographic analysis showed that the purified rCVN was an intact and homogeneous protein with a molecular weight of 11,016.68 Da. Potent antiviral activity of rCVN against herpes simplex virus type 1 and human immunodeficiency virus type 1/IIIB was confirmed in a dose-dependent manner at nanomolar concentrations. Thus, the His-SUMO double-fused CVN provides an efficient approach for the soluble expression of rCVN in the cytoplasm of E. coli, allowing an alternative system to develop bioprocess for the large-scale production of this antiviral candidate.

Introduction

Human immunodeficiency virus (HIV) remains a global health problem of unprecedented dimensions. Globally, there were an estimated 33 million people living with HIV (UNAIDS 2008). Cyanovirin-N (CVN), a potent anti-HIV protein shown promising in vivo efficacy recently (Buffa et al. 2009; Tsai et al. 2004), was originally discovered and isolated from the aqueous extract of the cyanobacterium (blue-green alga) Nostoc ellipsosporum (Boyd et al. 1997). CVN can permanently and efficiently inactivate an extensive range of HIV strains as well as prevent cell-to-cell fusion and the transmission of HIV. It acts by binding with high affinity to N-linked high-mannose glycans present on gp120 and gp41 (Dey et al. 2000; Shenoy et al. 2001), thus inhibiting the attachment and fusion of the virus particle to the target cell. In addition, CVN is also active against rhinoviruses, herpes simplex virus type 1 (HSV-1), hepatitis C, Ebola virus, and influenza A and B viruses (Barrientos et al. 2003; Helle et al. 2006; O'Keefe et al. 2003; Smee et al. 2008). Furthermore, CVN is insensitive to denaturants, detergents, organic solvents, and extreme temperatures. The potent HIV-inactivation ability of CVN and its physicochemical properties make this protein suitable for use as a topical anti-HIV drug.

CVN has a unique sequence of 101 amino acid residues (Boyd et al. 1997). Although the two halves (residues 1–50 and 51–101) in the monomeric CVN share a high degree of sequence homology, CVN has an extremely low sequence homology with any other known proteins. It has no homology greater than eight contiguous amino acids or 20% of the total sequence to any other known protein. In addition, it contains two intramolecular disulfide bonds (Bewley et al. 1998; Yang et al. 1999). These properties make the artificial production of this protein very difficult. The expression of recombinant CVN (rCVN) has been attempted in various species including Escherichia coli (Boyd et al. 1997; Colleluori et al. 2005; Mori et al. 1998), Pichia pastoris (Mori et al. 2002), and transgenic Nicotiana tabacum (Sexton et al. 2006), but the efficient expression of biologically functional rCVN is still problematic due to low yield, aggregation, and abnormal modifications. Thus far, E. coli is still considered to be the simplest, most efficient, and least expensive host to produce simple proteins such as rCVN (Liu and Schultz 2006). However, there is a complex reductase system in the cytoplasm of E. coli that renders it difficult to form stable disulfide bonds (Prinz et al. 1997), and these are crucial for the bioactivity of CVN. The periplasm is a more feasible location for the formation of disulfide bonds and protein refolding. But the yield is extremely low (usually less than 10 mg/L) for a heterologous protein produced in the periplasm (Mergulhao et al. 2004). In addition, when expressed in periplasm under the lead of the ompA signal peptide, the final rCVN product was a mixture of intact proteins: Up to 30% of the truncated product was missing the first two N-terminal residues (Mori et al. 1998; Prinz et al. 1997). Alternatively, the expression of CVN as inclusion bodies in the cytoplasm increased the yield to 3.5 mg rCVN/g of wet cell. Nevertheless, expression remains problematic as this method produces a significant fraction of dimeric species consistently, as well as generating uncharacterized side-chain modifications (Colleluori et al. 2005).

Small ubiquitin-related modifier (SUMO) family proteins function as posttranslational modifiers by attaching to other proteins covalently and reversibly (Saitoh and Hinchey 2000). Recently, SUMO, fused at the N-terminus with heterologous proteins, has been found to improve protein folding, enhance expression level, and protect the protein from degradation via its properties as a chaperonin. SUMO-fusion proteins can be specifically cleaved by SUMO protease depending on the 3D structure of the SUMO tag at the fusion site, and the target protein will be released, together with its native N terminus (Malakhov et al. 2004). Recently, a series of SUMO expression vectors have been developed for the efficient production of recombinant proteins in prokaryotic or eukaryotic host cells (Assadi-Porter et al. 2008; Liu et al. 2008; Sun et al. 2008; Ye et al. 2008).

Here, we tested the effect of Saccharomyces cerevisiae SUMO3 on the expression of rCVN in the cytoplasm of E. coli using a construct carrying sumo-fused cvn with a hexahistidine tag. We found that this new system produced soluble and biologically functional rCVN efficiently in the cytoplasm of E. coli, and this could be rapidly purified into intact and homologous native rCVN by two rounds of Ni²⁺ chelating chromatography intervened with a SUMO protease cleavage step. The low cost and reduced time consumption of this improved system makes it suitable for the large-scale production of rCVN.

Materials and methods

Gene splicing by overlap extension

A standard overlap extension was carried out using 20 pmol of each primer, 2 ng of template, 1 U of Pyrobest DNA polymerase (Invitrogen, Carlsbad, CA, USA) and other reagents as per the manufacturer’s instructions. Amplification (25 cycles) consisted of denaturation at 95°C for 1 min, annealing at 53°C for 45 s, and elongation at 72°C for 1 min (Horton et al. 1989).

Construction of the expression plasmid

The DNA sequence encoding SUMO was amplified from the vector, pET-SUMO, using the forward primer (F-SUMO, 5′-CAGCATATGCATCATCATCATC-3′) and the reverse primer (R-SUMO, 5′-CTGGGAGAATTTACCAAGACCACCAATCTGTT CTCTG-3′). The resulting 342-bp fragment was spliced with the encoding sequence of cvn with the primers of F-SUMO and C5 (5′-AGAGGATCCTCATCATTCGTATTT CAGGGTACCGTCGATGTTAGCGATGTGGTC GTC-3′). The open reading frame (ORF) of his6-sumo-cvn (642 bp) was then digested with NdeI and BamHI (Takara, Dalian, China), followed by subcloning into pET-3c (Novagen) to create the corresponding expression plasmid pET-SUMO-CVN, which was then transformed into E. coli DH5α (Invitrogen) to sequence the correct insert. At the same time, another expression plasmid, pET-CVN without SUMO, was also constructed using the same vector in the same site. All fragments of the DNA recombination were purified using agarose gel electrophoresis and the Gel Recovery Kit (Sangon, Shanghai, China).

Protein expression and fractionation

Individual colonies of E. coli BL21(DE3) (Novagen, Madison, WI, USA) harboring the pET-sumo-cvn (or pET-cvn) plasmids were inoculated in Luria–Bertani (LB) medium containing 100 μg/ml ampicillin. For optimal conditions, 0.5% glucose and 1.6 mM MgSO₄ were added to the medium. The cells were grown at 37°C, 30°C, or 20°C in a shaking incubator (180 rpm) until the culture reached an OD₆₀₀ = 0.6–1.0. At this point, 1 or 0.5 mM isopropyl-beta-d-thiogalactopyranoside (IPTG) was added, and the cells were grown for a further 4–24 h after induction. Cells were harvested by centrifugation at 4,000×g for 20 min and lysed in sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer for further analysis.

To determine the solubility of recombinant SUMO-CVN (rCVN), cells from 100 ml of induced culture were freeze-thawed once and thoroughly resuspended in NTA-10 buffer (20 mM Tris-HCl, pH 8.0, 0.5 mol/L NaCl, 10 mM imidazole) at a ratio of 1 g of cells to 5 ml of buffer, followed by the addition of 0.1 g/L lysozyme (Dingguo, China), and incubated at 30°C for 30 min. The whole cell lysate was then centrifuged at 20,000×g for 30 min at 4°C to separate the soluble and insoluble fractions.

Protein purification

A typical procedure for the purification of native rCVN is illustrated in Fig. 4a. The hexahistidine-tagged SUMO-CVN from E. coli BL21(DE3) was initially purified by ion metal affinity chromatography with Ni-NTA agarose (GE Healthcare, USA). Fifty milliliters of the cleared bacterial lysate from 1 L of LB culture was applied to an XK 16/20 column (GE Healthcare, USA) containing 20 ml of Ni-NTA resin that had been equilibrated with ten column volumes of NTA-10 buffer. The resin was then washed with NTA-10 buffer until the OD₂₈₀ reached a base line. The nonspecifically bound proteins were eluted with 100–150 ml of NTA-40 buffer (20 mM Tris-HCl, pH 8.0, 0.5 M NaCl, 40 mM imidazole). Finally, recombinant hexahistidine-tagged SUMO-CVN was eluted with NTA-200 buffer (20 mM Tris-HCl, pH 8.0, 0.5 mol/L NaCl, 200 mM imidazole).

Fig. 4.

Chemicophysical characterization of the recombinant native CVN. a Molecular weight of rCVN determined by MALDI-TOF MS. b Predicted peptide map of natural CVN. Amino acid sequences underlined with different colors indicate the nine fragments of trypsin-digested CVN. T1 (316.4) indicates that the peptide numbered T1 has a theoretical MW of 316.4 Da. The same labeling scheme applies for T2 through T9. c PMF analysis of rCVN. The rCVN band cut from a 16% Tris-tricine SDS-PAGE gel was digested with trypsin and then analyzed by MALDI-TOF-MS. Mass peaks that match the theoretical MW of fragments deriving from natural CVN are marked with asterisk. d Purity analysis of rCVN by RP-HPLC with C18 column

After being applied to a sephadex G-25 column (GE Healthcare, USA) and washed with buffer A (20 mM Tris-HCl, pH 8.0, 2% Igepal, 1.5 M NaCl, 10 mM dithiothreitol) to remove imidazole, the eluted SUMO-CVN was digested with SUMO protease (1 U enzyme to 10 μg substrate) at 4°C for 1 h (Lee et al. 2008). The Ni-NTA column was used again to remove SUMO, undigested SUMO-CVN, and SUMO protease (in which the hexahistidine tag was also fused to the N-terminal), thus resulting in the native rCVN. The purity of rCVN was analyzed on a 16% reduced Tris-tricine SDS-PAGE gel, and the protein concentration was determined using the Bradford method (Bradford 1976).

MALDI-TOF MS analysis

The purified native rCVN was subjected to matrix-assisted laser desorption ionization coupled to time-of-flight mass spectrometry (MALDI-TOF MS), to confirm its peptide mass map and molecular weight (MW; Henzel et al. 2003). Briefly, after separation on the 16% Tris-tricine SDS-PAGE gel, followed by silver staining, the rCVN band was excised and destained with 1% potassium ferricyanide and 1.6% sodium thiosulfate. The protein was then in-gel digested with 20 g/L trypsin at 37°C overnight. The digested peptides were extracted and loaded onto an MTP AnchorChip^TM 600/384 TF (Bruker-Daltonik, Bremen, Germany). MALDI-TOF MS analysis was carried out using an Ultraflex^TM MALDI-TOF mass spectrometer (Bruker-Daltonik). For molecular weight determination, the purified protein was loaded directly onto the palladium plate and analyzed with an Ultraflex^TM MALDI-TOF mass spectrometer.

RP-HPLC analysis

The purity of the purified rCVN was confirmed using reverse-phase high-performance liquid chromatography (RP-HPLC) in addition to the SDS-PAGE analysis. The sample (10 μL) was loaded onto a 4.6 × 250-mm Zorbax 300 SB-C18 column (Agilent Lifescience, USA) and then eluted with a linear gradient of 30–70% acetonitrile at a flow rate of 1 ml/min in the presence of 0.1% trifluoroacetic acid. The HPLC system (Agilent Lifescience, Model 1100, USA) was configured with a quarter pump and diode array detector.

MTT assay

The activity of rCVN against HSV-1 was determined using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) method (Takeuchi et al. 1991). Briefly, rCVN was twofold serially diluted from 181.55 nM (2 μg/ml) to 11.35 nM (0.125 μg/ml), and then 50 μL of each dilution and HSV-1 (100 TCID₅₀/50 μL; State Key Laboratory of Virology of Wuhan University, China) was added to the monolayer of Vero cells (ATCC, USA) in a 96-well microtiter plate. A cytotoxicity determination was simultaneously carried out in wells with the test samples but without the virus. Concurrently, 25 μg/ml acyclovir was used as a positive control. After incubation at 37°C with 5% CO₂ for 48 h, MTT reagent was added for color development. The absorbance of λ _570/630 was measured using a microplate reader (Bio-Rad Laboratories, USA) and the 50% inhibiting concentration (IC₅₀) and 50% cytotoxic concentration (CC₅₀) were calculated.

WST-1 assay

The anti-HIV activity of the recombinant CVN was determined in vitro using the water-soluble tetrazolium salt (WST)-1 method in a P3 laboratory (Boyd et al. 1997; Dueweke et al. 1993). CVN was twofold serially diluted with Rapid Prototyping and Manufacturing Institute 1640 medium (Invitrogen) containing 10% fetal bovine serum. Then 50 μL of each dilution and HIV-1/IIIB (100 TCID₅₀/50 μL, harvested from the culture supernatant of MOLT-4/IIIB cells), were added to MT-4 cells (10⁴ cells/100 μL/well) in 96-well microtiter plate. The final concentration of CVN ranged from 3.52 to 112.5 nM. At the same time, azidothymidine (AZT; Sigma) ranging from 0.98 to 250 nM was used as a positive control. After incubation with 5% CO₂ at 37°C for 96 h, WST-1 was added to develop the color. The absorbance of λ _450/650 was determined using a microplate reader (Wako, Japan) to evaluate the activity.

All samples were tested in triplicate, and the reproducibility was confirmed in three separate experiments.

Gel electrophoresis

SDS-PAGE (12%) or 16% tricine SDS-PAGE was carried out under reducing conditions as previously described (Laemmli 1970). The gel was scanned with a densitometer (Bio-Rad, USA) after staining with Coomassie Blue, and the expression level of rCVN was determined.

Results

Construction and soluble expression of SUMO-CVN

An artificial coding sequence for cvn was designed de novo, and nine oligos were synthesized for overlap extension of the coding sequence, according to the codon usage bias of E. coli (Fig. 1a). The sequence of his-sumo was amplified from pET3c-sumo using a primer pair in which a sequence coding for hexahistidine was added to the forward primer. The final product of the multiple overlap extension was cloned into pET-3c at the NdeI and BamHI restriction sites to get the expression plasmid pET-SUMO-CVN, which was then transformed into the expression host, E. coli BL21(DE3) for protein expression.

Fig. 1.

Construction and soluble expression of SUMO-CVN in the cytoplasm of E. coli. a Strategy for the synthesis of hexahistidine-tagged sumo-cvn ORF by multiple overlap extension. Nine primers were designed using the amino acid sequence of natural CVN and the codon bias of E. coli. Elongation of primer pair F1 and R1 resulted in the fragment C1 (97 bp). Taking C1 as a template and using F2 and R2 as primers, the fragment C2 (173 bp) was synthesized. A similar method was used to get the fragments: C3 (241 bp), C4 (298 bp), and C5 (cvn, 337 bp, overlapped with hexahis-sumo by 37 bp). The sequence of hexahis-sumo (342 bp) was amplified from pET3c-sumo using a primer pair in which the sequence corresponding to the hexahistidine tag was added in the forward primer, i.e., F-SUMO. In the final round of overlap extension, C5 and S1 were spliced into hexahis-sumo-cvn (642 bp) by the primer pair of F-sumo and R5. b Expression of SUMO-CVN in E. coli. c Expression of rCVN without SUMO in E. coli. E. coli BL21(DE3) harboring pET-sumo-cvn (b) or pET-cvn (c) was grown in LB medium to exponential phase (OD₆₀₀ = 0.6) and then induced with 1 mM IPTG for 4 h at 37°C. Cultures harboring an empty vector or without IPTG induction were used as negative controls. Whole cell lysates (wcl) were extracted, fractioned, resolved by SDS-PAGE, and stained with Coomassie Blue. Protein bands with expected sizes are indicated with arrows

After 4 h induction with 1 mM IPTG at 37°C, SUMO-CVN was efficiently expressed in the cytoplasm of E. coli to 17.5% of the total protein (Fig. 1b). Almost all of the target protein was found in the supernatant (up to 32.3% of the total soluble protein), suggesting that the protein was quite soluble. In contrast, expression of rCVN without the SUMO fusion in the same vector and host cell was not detected (Fig. 1c). Similar results were also obtained using a SUMO-negative CVN in pBV220 in various E. coli strains including BL21(DE3), Origami(DE3), JM109, and DH5α (data not shown). pBV220 is an expression vector with a strong promoter that we have successfully used to express hNDPK-A previously (Xiong et al. 2007). We also tested the possibility of expressing SUMO-negative rCVN in the periplasm of E. coli instead of the cytoplasm and still found no visible expression. Thus, we have showed for the first time that SUMO can effectively stabilize and facilitate the soluble expression of rCVN in the cytoplasm of E. coli.

Purification of native rCVN from the SUMO-CVN fusion

Before purification of native rCVN from the SUMO-CVN fusion protein, we first optimized the expression conditions for the target protein in E. coli BL21(DE3) by testing various culture temperatures, induction periods, and IPTG concentrations (Fig. 2). No significant difference in the expression of target protein was found when either 0.5 or 1 mM IPTG was used at induction. For cultures induced at 20°C, 30°C, and 37°C for 4 h, the expression of SUMO-CVN increased gradually with increasing temperature. Longer induction periods up to 24 h further increased the expression level, and the highest expression level was obtained in cultures induced with 0.5 mM IPTG at 20°C for 24 h. At these optimum conditions, the expression level was 28.3 ± 0.56% of the total protein.

Fig. 2.

Optimization of the expression of SUMO-CVN in E. coli BL21(DE3) under different induction conditions. Expression of SUMO-CVN was induced at various culture temperatures, induction periods, and IPTG concentrations. Whole cell lysates were resolved in 12% SDS-PAGE gels and stained with Coomassie Blue. a A representative gel image is shown here and the arrow indicates SUMO-CVN, whose apparent molecular weight is approximately 28 kDa. The expression level of SUMO-CVN shown in a was determined by densitometer scanning. b Data are presented as means ± standard deviation (n = 3)

After induction with IPTG under the optimum conditions, the total soluble protein was extracted and applied to the Ni-NTA column. After washing with buffer containing 200 mM imidazole, SUMO-CVN was subsequently eluted with a purity of 73.8% (Fig. 3b) and then digested with SUMO protease to release the native rCVN protein. By running the cleaved proteins through Ni-NTA column again to capture hexahistidine-tagged proteins, native rCVN could be separated from SUMO, undigested SUMO-CVN, and SUMO protease. Approximately 3 to 4 mg of native rCVN could be purified from 1 g of wet cells using this process, to a purity up to 97.6%. This demonstrates that a considerable amount of native rCVN protein could be easily and economically purified with this system, making it suitable for future large-scale production of this protein.

Fig. 3.

Purification of native CVN from the hexahistidine-tagged SUMO-CVN. a Flow chart of the purification process. Native rCVN was purified by two rounds of Ni²⁺ chelating chromatography, intervened with one round of hexahistidine-tagged SUMO protease cleavage. b SDS-PAGE analysis of purified rCVN. pET-sumo-cvn bearing E. coli BL21(DE3) was cultured in medium containing 0.5 mM IPTG at 20°C for 24 h. Fractions from the purification were collected and resolved using SDS-PAGE. Lane 1 eluted fraction of the first Ni²⁺-chelating chromatography; lane 2 SUMO protease cleaved mixture (single asterisk indicates the released hexahistidine-tagged SUMO, double asterisks indicates the rCVN); lane 3 hexahistidine-tagged SUMO protease; lane 4 the eluted fraction of the second Ni²⁺-chelating chromatography; lane 5 low-molecular-weight protein markers

Chemicophysical characterization of the recombinant native CVN

In addition to the formation of inclusion bodies, other unresolved problems for the expression of rCVN in E. coli include N-terminal deletion, dimerization, and side-chain modification, resulting in different apparent MWs of the final recombinant product (Mori et al. 1998). To confirm the structural integrity of rCVN expressed from our system, we determined the MW of the purified rCVN using MALDI-TOF MS. It was found that the MW of the rCVN was 11,016.68 Da, in agreement with the theoretical MW for natural CVN (Fig. 4a). There are eight known trypsin cleavage sites in natural CVN, making it possible to confirm the sequence identity of the rCVN product. We determined the peptide profile of rCVN by peptide mass fingerprint (PMF) analysis. The predicted cleavage map of natural CVN is shown in Fig. 4b. From MALDI-TOF/TOF MS spectra of in-gel digested rCVN (Fig. 4c), six peptide fragments with molecular weights comparative to their natural counterpart were identified. Database interpretation also confirmed the identity of the rCVN. Purity is another important characteristic for a recombinant drug candidate. In addition to SDS-PAGE analysis (Fig. 3b), RP-HPLC was also employed to determine the purity of the final product. Figure 4d showed that the elution profile of rCVN in a C18 column is a smooth Gaussian curve without any shoulder peak. Taken together, these results provide evidence that the SUMO expression system we used is able to produce high-quality rCVN.

Antiviral activity of recombinant native CVN

Finally, we examined the biological activity of purified rCVN, using the RNA virus HIV-1/III B and the DNA virus HSV-1 as models to evaluate the antiviral activity of rCVN. A WST-1 assay revealed that native rCVN possesses significant antiviral activity. For HIV-1/IIIB, the IC₅₀ and CC₅₀ were 22.35 ± 3.74 and 164.31 ± 5.90 nM, respectively (Table 1). At concentrations of 56.25 and 28.13 nM, the inhibition rates of rCVN were 71.2% and 62.8%, respectively. The control, AZT, a well-known anti-HIV compound, exhibited 64.62% inhibition of the virus at a concentration of 62.5 nM (Fig. 5). The IC₅₀ of rCVN for HSV-1 was 28.14 ± 2.72 nM, and the CC₅₀ was 190.63 ± 9.07 nM (Table 1). The above results indicate that purified rCVN is functionally active. In addition, SUMO-CVN also exhibited activity against HSV-1 with an IC₅₀ of 31.37 ± 2.15 nM, comparable to that of native rCVN. This result serves as evidence that SUMO-CVN is produced in the cytoplasm as the fully folded and biologically functional form, providing an alternative source of the protein for practical applications.

Table 1.

Summary of IC₅₀ and CC₅₀ of native rCVN against HIV-1/IIIB and HSV-1

Viral strains	Tested samples	IC50 (nM)	CC50 (nM)
HIV-1/III B	rCVN	22.35 ± 3.74	164.31 ± 5.90
	AZTa	35.88 ± 4.30	>2,500
HSV-1	rCVN	28.14 ± 2.72	190.63 ± 9.07
	SUMO-CVN	31.37 ± 2.15	ND

IC₅₀ 50% inhibitory concentration, CC₅₀ 50% cytotoxic concentration, ND not done, AZT azidothymidine

^aAZT was used as a positive control for the anti-HIV-1/IIIB assay

Fig. 5.

Inhibition effect of rCVN on HIV-1/IIIB. The anti-HIV-1/IIB activity of rCVN was evaluated by WST-1 assay. Data are presented as means ± standard deviation from three independent experiments in which all of the samples were tested in triplicate

Discussion

Large-scale production of rCVN as a potential antiviral drug using genetically engineered microorganisms is required to provide a ready source of material for practical application (Fischetti et al. 2009; Sexton et al. 2006). However, the sequence properties of CVN make it difficult to be produced in heterologous hosts. The expression of rCVN in transgenic plants is time-consuming, with low yields, undeveloped downstream processes, and potential posttranslational modifications (Sexton et al. 2006). Another choice is to produce rCVN in a different eukaryotic system, P. pastoris. However, it seems difficult to obtain native rCVN using this system because of high-mannose N-glycosylation at position N30 and the dimerization of the final product mediated by P51 in the hinge region, both of which result in a loss of antiviral activity of rCVN (Mori et al. 2002). Although E. coli is so far the best-studied host, considerable problems still exist in its practical use for the expression of rCVN. Low protein yield remains a major issue. In addition, rCVN as expressed in E. coli tends to aggregate into nonfunctional forms, which have been attributed to the lack of stable disulfide bond formation and proper folding.

In an attempt to solve the above problems, we fused CVN with yeast SUMO3, preceded by a hexahistidine tag, and we studied the efficiency of this system at expressing rCVN in E. coli. SUMO is a common protein in eukaryotic cells, and it regulates the activities and cellular localizations of target proteins through a modification process similar to ubiquitination (Butt et al. 2005; Kim et al. 2002). Several proteins that are difficult to express in conventional systems, such as SARS-CoV structure proteins (Zuo et al. 2005), metalloprotease (MMP13) (Marblestone et al. 2006), brazzein (Assadi-Porter et al. 2008), urodilatin (Sun et al. 2008), and antibody fragment (Ye et al. 2008) have been successfully expressed in E. coli using a SUMO fusion strategy. Our data showed that a SUMO-CVN fusion construct was efficiently expressed in a soluble form in the cytoplasm of E. coli, supporting the idea that SUMO facilitates the soluble expression of foreign (heterologous) proteins in E. coli. As SUMO is a fusion label, the mechanism of the yield enhancement and solubility improvement is still unclear. Several studies implicate that it may function as a chaperonin or act as a core for the folding of recombinant proteins in E. coli (Butt et al. 2005).

As there is no specific affinity method available, native CVN is usually isolated by general chromatography techniques such as RP-HPLC and anion exchange. Ni²⁺ affinity chromatography has also been applied to rapidly purify rCVN produced in P. pastoris (Mori et al. 2002), but the final product is His-tagged rather than native rCVN. It is possible to release native rCVN from SUMO-CVN by cleaving with SUMO protease at the site of the fusion. An advantage of our system is that a hexahistidine tag was fused at the N-terminal of SUMO so that the SUMO-CVN could be rapidly purified from the total protein extraction by simply running it through a Ni-NTA column. After digestion with SUMO protease, the released native rCVN could be subsequently purified using the same column to over 97% purity. With this purification process, we obtained a final yield of about 3–4 mg/g of wet cells. The amount is considerable as the protein was induced in E. coli in a conventional flask-shaking culture. E. coli has been cultured to densities greater than 150 g DCW/L in optimized conditions (Lee 1996; Xia et al. 2008). In an ordinary fermentor culture, the biomass will be over 50 g WCW/L, a level that is usually reached in our laboratory (Xiong et al. 2007). The yield of rCVN from our SUMO-CVN system in a fermentor fed-batch culture is therefore expected to be over 150 mg/L. Efforts are now being made to further improve the protein yield of native rCVN using our system.

The expression of soluble SUMO-CVN is a major advantage of our expression system. The highest yield of rCVN reported thus far is 140 mg/L, which has been achieved using a fermentor culture (Colleluori et al. 2005). However, the protein was expressed as inclusion bodies, which dramatically decreased protein quality. Other problems reported for rCVN expression in E. coli include N-terminal deletion, dimerization, and side-chain modification. MW determination, PMF analysis, RP-HPLC chromatography, and SDS-PAGE demonstrated that the purified rCVN derived from our system was intact, homogeneous, and monomeric. Another advantage of this system is that the intermediate recombinant product, SUMO-CVN, was also biologically functional and had an antiviral activity against HSV that was comparable to native rCVN. This result demonstrates that the SUMO-CVN produced in the cytoplasm of E. coli is fully folded and in its functional form, providing an alternative source for practical applications.

Comparing the IC₅₀ and CC₅₀ of rCVN against HIV-1 with that of the natural one (Boyd et al. 1997), it looks like that the rCVN is not as potent as the natural product because the IC₅₀ reported here is significant higher than the previous report (0.4 nM). This might have resulted from the different bioassay parameters such as the host cell, the origination of virus and its titer in the detection system, or from the trace impurity in the recombinant product. The data of commercialized AZT also displayed higher IC₅₀ than the previous report (6 nm; Mitsuya et al. 1985), suggesting a system difference in our bioassay method.

CVN is a promising candidate for a topical HIV microbicide. A 5-mg dose, which proved to be effective in macaque studies, requires a production capacity of 5,000 kg a year to supply just ten million women twice a week (Shattock and Moore 2003). This scale of protein production may be achievable only by expression system that could produce rCVN with high productivity and low cost. Here, we propose an improved approach that efficiently expresses rCVN in the cytoplasm of E. coli. By fusing rCVN with SUMO, it can be efficiently produced in the cytoplasm of E. coli at a level of >30% of the total soluble protein. With the addition of a hexahistidine tag fused at the N-terminal of SUMO, intact and native rCVN can be rapidly purified in a soluble and biologically active form. Further improvement is still required to eventually bring this promising antiviral candidate from bench to clinical application.