Recombinant production of Zymomonas mobilis pyruvate decarboxylase in the haloarchaeon Haloferax volcanii

The unusual physiological properties of archaea (e.g., growth in extreme salt concentration, temperature and pH) make them ideal platforms for metabolic engineering. Towards the ultimate goal of modifying an archaeon to produce bioethanol or other useful products, the pyruvate decarboxylase gene of Zymomonas mobilis (Zm pdc) was expressed in Haloferax volcanii. This gene has been used successfully to channel pyruvate to ethanol in various Gram-negative bacteria, including Escherichia coli. Although the ionic strength of the H. volcanii cytosol differs over 15-fold from that of E. coli, gel filtration and circular dichroism revealed no difference in secondary structure between the ZmPDC protein isolated from either of these hosts. Like the E. coli purified enzyme, ZmPDC from H. volcanii catalyzed the nonoxidative decarboxylation of pyruvate. A decrease in the amount of soluble ZmPDC protein was detected as H. volcanii transitioned from log phase to late stationary phase that was inversely proportional to the amount of pdc-specific mRNA. Based on these results, proteins from non-halophilic organisms can be actively synthesized in haloarchaea; however, post-transcriptional mechanisms present in stationary phase appear to limit the amount of recombinant protein expressed.


Introduction
Archaea encompass a diversity of organisms able to grow in a wide range of environments.Temperatures > 100 °C (Blochl et al. 1997, Kashefi andLovley 2003), pH < 1 (Schleper et al. 1995, van de Vossenberg et al. 1998) and salt concentrations > 5 M (Martin et al. 2000) are just a few of the extremes in habitat where archaea are found.Although this diversity provides tremendous opportunities in metabolic engineering, it remains a relatively untapped source for such objectives (Schiraldi et al. 2003).Halophiles, hyperthermophiles, acidophiles and methanogens are ideal candidates for metabolic engineering because of their ability to survive in extreme environments and their unique metabolic pathways (Herbert 1992, Schiraldi et al. 2003).Potential applications of archaea in biotechnology are far-reaching, ranging from bioremediation of toxic compounds to production of biopolymers with medicinal applications (Bertrand et al. 1990, Emerson et al. 1994, Hezayen et al. 2000, Margesin and Schinner 2001a, 2001b, Schiraldi et al. 2003).Archaea are presently employed in several industrial processes, including wastewater treatment, fermentation of food products and production of solar salt (Thongthai et al. 1992, Scherer et al. 2000, Javor 2002).As a group, however, the domain is still very much underrepresented in metabolic engineering.Current applications of archaea in biotechnology typically incorporate the properties of the native organism but do not use engineered strains for optimization of product formation.
Halophilic archaea are especially well suited for biotechnological applications because of established genetic exchange systems and the large amount of biomass attainable from these organisms (Blaseio and Pfeifer 1990, Holmes et al. 1994, Ng et al. 2000).Unique features of the haloarchaea also make them attractive candidates for industrial use, including cell lysis upon exposure to low salt concentrations, which reduces the cost of protein purification (Ventosa et al. 1998).In addition, haloarchaea are easily cultured with little risk of contamination due to the high salt concentrations in which they grow, resulting in easier maintenance of cultures (Margesin and Schinner 2001b).
This study was undertaken to produce a non-halophilic metabolic enzyme in a haloarchaeal host as part of an endeavor to better utilize this diverse group of organisms in biotechnology.Specifically, the pyruvate decarboxylase gene of Zymomonas mobilis was expressed in recombinant Haloferax volcanii with the long-term goal of engineering archaeal metabolism through recombinant technologies.Pyruvate decarboxylase (PDC) was employed based on its application in the industrial production of biofuels and (R)-phenylacetylcarbinol (Iding et al. 1998, MacDonald et al. 2001, Wyman 2001) and limited distribution in nature (Candy andDuggleby 1998, Raj et al. 2002).Currently, commercially viable, large-scale production of PDC-based products from lignocellulosic materials via simultaneous saccharification and fermentation (SSF) is restricted (Ingram et al 1999).Engineering PDC production in robust microorganisms such as H. volcanii, which can use a variety of sugars (Rawal et al 1988) and withstand adverse environmental conditions (e.g., elevated temperature and high salt concentration), is expected to advance this area of biotechnology.

Materials
Biochemicals were purchased from Sigma-Aldrich (St. Louis, MO).Other organic and inorganic analytical grade chemicals were from Fisher Scientific (Atlanta, GA).Restriction endonucleases and DNA-modifying enzymes were from New England BioLabs (Beverly, MA).SnakeSkin dialysis tubing was from Pierce (Rockford, IL).Hybond-P membranes used for immunoblot were from Amersham Pharmacia Biotech (Piscataway, NJ).

Plasmids and cloning
Plasmids used in this study are summarized in Table 1.For expression of ZmPDC in H. volcanii, the Zm pdc gene was isolated from the pET21d-based plasmid pJAM431 by restriction enzyme digestion (XbaI and DraIII) and blunt-end ligated with the 9.9-kb BamHI-to-KpnI fragment of plasmid pBAP5010.The DNA fragment of pJAM431 used for ligation included not only the Zm pdc gene but also the Shine-Dalgarno sequence and T7 terminator of the original pET21d

Purification of ZmPDC
Protein purification steps were performed at room temperature unless otherwise indicated.Buffers were supplemented with 1 mM TPP, 1 mM MgCl 2 and 1mM dTT.Centrifugations were at 16,000 g (20-30 min, 4 °C).Cells were harvested by centrifugation and stored at -70 °C.Cells were resuspended in 2.5 to 6 volumes (wet w/v) of 50 mM NaPO 4 buffer at pH 6.5 (Buffer A) and lysed by passage through a chilled French pressure cell at 20,000 psi followed by centrifugation.Dialysis was at 4 °C for 16 h (< 1 ml of sample against 3 l of buffer) followed by centrifugation.Samples were filtered (0.45-µm) before column chromatography.
(i) Purification from recombinant E. coli (pJAM431) Cells of E. coli BL21(DE3) were freshly transformed with pJAM431.Cells (1 l) were grown to log phase (OD 600 of 0.6 to 0.8), and pdc gene expression was induced with isopropyl-β-D-thiogalactopyranoside (1 mM) for 2 h.Cell lysate was subjected to thermal treatment (60 °C, 30 min).The sample was applied to a Q Sepharose Fast Flow 26/10 column (Pharmacia) equilibrated in Buffer A and eluted into Buffer A supplemented with 1 M NaCl.Fractions with PDC activity (1 M NaCl) were pooled and directly applied to a 5-ml Bio-scale hydroxyapatite type I column (Bio-Rad, Hercules, CA) equilibrated in Buffer A. ZmPDC was eluted with a linear gradient of 50 to 500 mM sodium phosphate buffer at pH 6.5 in 50 ml.Fractions with PDC activity (350-500 mM sodium phosphate) were pooled and applied to a Superdex 200 HR 10/30 column (Pharmacia) equilibrated in Buffer A. Column eluant with PDC activity was pooled, stored at 4 °C and used for biochemical and structural analyses.
(ii) Purification from recombinant H. volcanii (pJAM206) Haloferax volcanii (pJAM206) (10 l) was grown to log phase.Cell lysate was applied to a 5-ml Bio-scale hydroxyapatite type I column (Bio-Rad) equilibrated in Buffer A. ZmPDC was eluted with a linear gradient of 50 to 500 mM sodium phosphate at pH 6.5 in 85 ml.Hydroxyapatite fractions with PDC activity (250-450 mM sodium phosphate) were pooled and dialyzed against Buffer A supplemented with 1 M (NH 4 ) 2 SO 4 .Sample was applied to a 1 ml Phenyl-Sepharose FF (low sub) column (Pharmacia) and eluted with a linear gradient of 650 mM to 0 M (NH 4 ) 2 SO 4 in 20 ml Buffer A. Fractions with PDC activity were concentrated with a Centricon 10 (Amicon, Billerica, MA).Sample (1 mg at 2.25 mg ml -1 ) was applied to a Superdex 200 HR 10/30 column (Pharmacia) equilibrated in Buffer A. Fractions with PDC activity were applied to a 1-ml DEAE-cellulose column (Sigma) equilibrated in Buffer A. Column eluant with PDC activity was pooled, stored at 4 °C and used for biochemical and structural analyses.

Protein analysis and enzymatic assays
Protein concentrations were determined by the Coomassie blue dye-binding assay with bovine serum albumin as the standard (Bio-Rad) (Bradford 1976).ZmPDC activity was monitored in a 1-ml coupled assay, which measured the pyruvatedependent oxidation of NADH with yeast alcohol dehydrogenase (ADH) as described previously (Conway et al. 1987).One unit of activity (U) is defined as the amount of PDC that generates 1 µmol of acetaldehyde per min.Alcohol dehydrogenase activity was measured as the NAD-dependent oxidation of ethanol (0.05% (v/v)) as described previously (Neale et al. 1986).Yeast ADH was included as a positive control.

Circular dichroism
Circular dichroic spectra were obtained on a Jasco 500C spectropolarimeter (Easton, MD) at varying temperatures with a 1-mm path length, water-jacketed quartz cuvette.Instrument parameters were set to 1.0-nm bandwidth, 50-nm min -1 scan rate and 0.2-nm resolution.Instrument calibration was performed with 0.06% (w/v) (1S)-(+)-10-camphorsulfonic acid and neodymium glass (Hennessey and Johnson 1982).Chicken egg white lysozyme at 7 µM concentration in 5 mM NaPO 4 buffer at pH 6.5 was used as a standard.The spectra of ZmPDC (100 to 160 nM) dialyzed into 5 mM NaPO 4 buffer at pH 6.5 were obtained from 250 to 195 nm.A mean of 10 accumulations was generated with the sample spectra taken for a minimum of three separate runs.Samples were allowed to equilibrate (5 min) at the set temperature before measurement.Noise reduction of spectra was performed with the Jasco J700 software package (Version 1.50.01).Secondary structure content of proteins was calculated from circular dichroism spectra as described previously (Chen et al. 1974, Deleage and Geourjon 1993, Sreerama and Woody 1993).α-Helical content was used for the comparison of the secondary structures, since current deconvolution methods are most reliable in the identification of α-helices compared with β-sheets and random coils (Wallace 2000).

Quantitative RT-PCR analysis
Total RNA was purified from H. volcanii DS70 (pJAM206) ARCHAEA ONLINE at http://archaea.wsPYRUVATE DECARBOXYLASE IN RECOMBINANT H. VOLCANII with the RNeasy MiniKit (Qiagen) protocol for the isolation of bacterial RNA with the following modifications.Cells were centrifuged (1 min at 18,000 g) and immediately resuspended in 100 µl RNase-free water.An additional DNase treatment (Deoxyribonuclease I, amplification grade) was performed after RNA purification according to the supplier (Sigma).The total RNA concentration was determined by absorbance at 260 nm.Ribosomal RNA detected by ethidium bromide stain after gel electrophoresis served as the invariant control between total RNA samples prepared from log and stationary phase cells (OD 600 of 1.1 and 3.0, respectively).One-step quantitative RT-PCR was performed with the one step Quantitect SYBR Green RT-PCR Kit (Qiagen) and iCycler MyiQ Real-Time PCR Detection System (Bio-Rad).A dilution series of total RNA served as the template with the oligonucleotide primers 5′-TGGCGAACTGGCAGAAGCTATCA-3′ and 5′-CGCGCTTACCCCATTTGACCA-3′.The reference standard was Z. mobilis pdc-specific mRNA generated by in vitro transcription with the HindIII DNA fragment of plasmid pJAM431 and the MAXIscript In Vitro Transcription Kit (Ambion, Austin, TX).Reverse transcriptase-PCR included a reverse transcription step (50 °C, 30 min) and reverse transcriptase inactivation and DNA polymerase activation step (95 °C, 15 min).This was followed by 40 cycles of denaturation at 94 °C (15 s), annealing at 67 °C (30 s) and extension at 72 °C (30 s).A negative control incubated on ice (versus 50 °C) for the reverse transcriptase step was included to ensure that total RNA was not contaminated with genomic DNA.Data were analyzed with the MyiQ Single-Color Real-Time PCR Detection System software Version 1.0 (Bio-Rad).Reverse transcriptase-PCR was performed in triplicate with amplification efficiencies between 90 and 100%, and melt curve analysis was performed to verify uniformity of the product.

Analysis of ethanol and organic acid production
Ethanol production was monitored by gas chromatography as described previously (Beall et al. 1991).Ethanol standards ranging from 0.1 to 10 g l -1 , 0.05% (v/v) formaldehyde and 1% (v/v) isopropanol comprised the internal standard.Sterile medium supplemented with 8% (w/v) ethanol was included as a control to confirm that less than 10% of the ethanol volatilized during incubation at 37 °C for 5 days.Organic acids and glucose were measured as described previously (Causey et al. 2003) with a Hewlett-Packard HPLC (HP 1090 series II, Palo Alto, CA) equipped with a UV monitor (210 nm) and refractive index detector.Products were separated with a Bio-Rad HPX-87H column (10 µl injection) with 4 mM H 2 SO 4 as the mobile phase (0.4 ml min -1 , 45 °C).

ZmPDC is active in high salt concentrations
Previous studies revealed that ZmPDC is relatively stable and active at high temperatures (50-60 °C) (Conway et al. 1987, Raj et al. 2002).Thus, the structural forces that stabilize ZmPDC at these high temperatures may also stabilize the enzyme at high salt concentrations (Keradjopoulos and Holldorf 1977).To investigate this, ZmPDC was purified from recombinant E. coli (ZmPDC Eco ) and assayed at increasing concentrations of NaCl and KCl (Figure 1).The enzyme retained nearly 50% activity when assayed in buffer supplemented with either 2 M NaCl or KCl.Furthermore, PDC retained 20% activity when assayed in the presence of 4 M salt.Pre-incubation of ZmPDC for 30 min in the salts prior to assay did not influence this relative activity.Based on these results, the ZmPDC is quite active at high salt concentrations and thus is an ideal candidate for channeling pyruvate to acetaldehyde in the high ionic strength cytosol of a haloarchaeon such as H. volcanii.

ZmPDC produced in recombinant H. volcanii is active
To engineer H. volcanii for ZmPDC production, the pdc gene was expressed from a plasmid (pJAM206) which included the ribosomal RNA P2 promoter from Halobacterium cutirubrum and T7 terminator from a pET plasmid to control transcription (Table 1).Immunoanalysis of log-phase H. volcanii (pJAM206) cells using anti-ZmPDC antibody revealed a cross-reacting protein of 60 kDa that was absent in the parent strain (data not shown).This 60-kDa protein corresponded in molecular mass to ZmPDC.A low, but detectable, level of PDC activity (0.12 U per mg protein) was present in H. volcanii (pJAM206) cell lysate but was not detected in the parent strain.These results indicated that the ZmPDC protein was produced and was active.The expression level was estimated to be 0.1% of soluble protein based on the specific activity of the cell lysate.

ZmPDC from H. volcanii is active and comparable in structure to that from E. coli
It has been shown previously that the ZmPDC purified from recombinant E. coli (ZmPDC Eco ) is comparable in biochemical properties to that of the native ZmPDC (Hoppner andDoelle 1983, Neale et al. 1987).Because of the ease of expres-Figure 1.Effect of salt on the pyruvate decarboxylase (PDC) activity of Z. mobilis purified from recombinant E. coli (ZmPDC Eco ).Purified enzyme was incubated (15 min, 25 °C) and assayed in buffer supplemented with NaCl (᭜) and KCl (ᮀ) at the concentrations indicated.sion and purification from E. coli, the ZmPDC Eco was used for comparison with ZmPDC purified from H. volcanii (ZmPDC Hvo ) (Figure 2).The ZmPDC Hvo had 55 U mg -1 of activity compared to 91 U mg -1 of activity for the ZmPDC Eco (Table 2).Both proteins purified as tetrameric complexes (240 kDa) based on gel filtration.Circular dichroism revealed no major differences in secondary structure between the ZmPDC purified from either organism (Figure 3).The circular dichroism spectra were identical with both exhibiting the characteristic minima at 208-209 nm and 222 nm and maximum at 195 nm consistent with a high degree of α-helical structure.The deconvolution methods utilized reflect this uniformity between the spectra, indicating little difference in the overall α-helical content.Additionally, the predicted α-helical content from the circular dichroism measurements (36-42.2%)correlates well with the published crystal structure estimate of 36.8% for ZmPDC Eco (Dobritzsch et al. 1998).Together these results reveal similar secondary structure between the ZmPDC Hvo and ZmPDC Eco .
The reason for the reduced (40%) activity of ZmPDC Hvo compared with ZmPDC Eco remains to be determined.The similarity of protein structure in the two preparations suggests that no perturbations to overall structure occurred when ZmPDC was expressed in H. volcanii.Instead, the 460-fold purification required for a homogeneous preparation of ZmPDC Hvo compared with the 12.4-fold purification for ZmPDC Eco (Table 2) may have contributed to a loss of the thiamine pyrophosphate (TPP) cofactor.Such a tetrameric form of ZmPDC, which retains its overall configuration but has lost cofactors and activity, has been reported (Candy et al. 1996).In fact, cofactor loss is a common problem when purifying TPP-dependent enzymes, which is why cofactors are typically included in all purification buffers and the pH is maintained below neutrality.Thus, although the ZmPDC purified from H. volcanii as a tetrameric complex, it does not preclude the possibility that a fraction of this preparation lacks the necessary cofactors.

Amount of soluble ZmPDC varies with the growth phase of H. volcanii (pJAM206)
To better understand and enhance expression of ZmPDC, the amounts of this foreign protein and its mRNA were monitored throughout growth of H. volcanii (pJAM206) on rich medium.The ZmPDC protein was detected in log phase and in the transition to the stationary phase but not in the stationary phase cells (Figure 4).Although transcription from rRNA promoters often declines in stationary phase (Aviv et al. 1996, Rosado andGage 2003), the percentage of Zm pdc-specific mRNA to total RNA was actually threefold higher at this phase of growth (1.38 ± 0.19%) than at log phase (0.42 ± 0.05%).These results reveal a growth-phase-dependent decrease in the amount of ZmPDC protein that was inversely proportional to Zm pdc-specific mRNA.Thus, the reduced amount of ZmPDC protein in the stationary phase was not due to a lack of mRNA transcript.Instead, factors beyond transcription were responsible.
A variety of post-transcriptional factors may be responsible for the low amounts of ZmPDC protein in H. volcanii, includ-ing proteolysis, inclusion body formation and reduced translation of the pdc-specific transcript.Our previous work indicates that differences in codon usage between pdc genes and the host organism can severely impair overall recombinant protein production (Talarico et al. 2001).Examination of the differences between the Zm pdc gene and H. volcanii genome suggest this may contribute to the low quantity of ZmPDC in this recombinant organism.The Zm pdc gene makes increased use of the codons: GCU, AAU, CAU, CCU, UAU and GUU compared with the H. volcanii genome.It is also possible that proteasome-mediated degradation is responsible for the low quantities of ZmPDC in stationary phase H. volcanii.Recent work indicates that the PanB and α2 proteins of this proteolytic system increase several-fold during the transition to stationary phase (Reuter et al. 2004).(Tomlinson and Koch 1974, Rawal et al. 1988, Oren and Gurevich 1994aand 1994b).Breakdown of glucose follows a modified Entner-Doudoroff pathway in which glucose is first oxidized to gluconate.The phosphorylation step acts on 2-keto-3-deoxygluconate to form 2-keto-3-deoxy-6-phosphogluconate (KDPG), which is cleaved into glyceraldehyde-3phosphate and pyruvate.
Oxidoreductases that catalyze the reduction of acetaldehyde to ethanol are widespread in nature (Jez et al. 1997) and have been identified in thermophilic and hyperthermophilic archaea (Radianingtyas and Wright 2003).Although ADH activity has not been reported in the haloarchaea, the partial genome sequence of H. volcanii reveals putative Zn-dependent ADHs (e.g., GenBank Accession no.T44975).Thus, it was possible that the recombinant H. volcanii strain expressing PDC would generate ethanol.Ethanol, however, was not detected in the spent medium of either parent or recombinant H. volcanii strains (Table 3).In contrast, ethanologenic strains of E. coli produced 0.34-3.1 g ethanol l -1 (Table 3).Further analyses of the H. volcanii strains revealed acetate and lactate were the major organic acids produced with no detectable amounts of ADH activity in cell lysates (data not shown).Thus, ethanol production may be limited not only by the amount of PDC but also by ADH in H. volcanii (pJAM206).

Conclusions
This study demonstrates that non-halophilic proteins can be expressed and purified from recombinant haloarchaea in a form that is active and has similar structure to that purified from non-halophilic sources.The synthesis and purification of an active non-halophilic bacterial protein from H. volcanii demonstrates the potential capabilities of genetic engineering archaea as well as halophiles, thus expanding the possibility for production of novel metabolites in these microorganisms.Future endeavors focused on the metabolic engineering of archaea will allow for a greater diversity of platform organisms to produce novel metabolites and improve production efficiencies of current industrial processes.H. volcanii (pJAM206).The ZmPDC protein was detected in cell lysate (65 µg) by immunoblot.Haloferax volcanii DS70 grown to log phase was included as a control (Lane 1).Haloferax volcanii (pJAM206) was grown to log phase (Lane 2), the transition to stationary phase (Lane 3) and stationary phase (Lane 4) (OD 600 of 0.848, 2.06 and 3.27, respectively).Table 3. Ethanol production in recombinant E. coli and H. volcanii strains.Abbreviation: UD = undetectable (< 0.1 g l -1 ).
The start codon of the modified pdc gene was positioned 75 bp downstream of the Halobacterium cutirubrum rRNA P2 promoter.This resulted in the generation of shuttle expression plasmid pJAM206 for the synthesis of ZmPDC in H. volcanii.

Table 2 .
Purification of Z. mobilis pyruvate decarboxylase from recombinant H. volcanii and E. coli strains.
Pyruvate serves as a central intermediate in the metabolism of biomass and can be converted to a variety of end products, including ethanol.Pyruvate decarboxylase is central in channeling pyruvate to acetaldehyde, which is reduced to ethanol by ADH.Species of haloarchaea, including H. volcanii, are able to generate a pyruvate pool via the metabolism of glycerol and sugars