BACKGROUND OF THE INVENTION

Brazzein encompasses a group of related sweet proteins extracted and purified from the fruit of the African plant, Pentadiplandra brazzeana. (Ming and Hellekant, FEBS Lett. 355: 106-108, 1994). Brazzein occurs in three naturally occurring forms, the amino acid sequences of which are known : Type I is 54 amino acids in length and has an N-terminal pyrrolidone carboxylic acid; Type II is 54 amino acids in length and has an N-terminal glutamine, and Type III is 53 amino acids in length and has an N-terminal aspartic acid ( Assad i- Porter, et al., Arch. Biochem. Biophys. 376: 252-258, 2000). In addition to the natural forms of brazzein, variants of brazzein, i.e., mutated and recombinant proteins with varied sequences and properties, have been created by genetic engineering, such as those described in U.S. Pat. Nos. 5,326,580; 5,346,998; 5,527,555; 5,741,537; 7,153,535; 6,274,707 and in Jin, et al., Chemical Senses 28: 491-498; Assad i- Porter, et al., Arch. Biochem. Biophys. 376: 259-265, 2000; Assadi- Porter, et al. , Chem. Senses 30 (Suppl. 1) : i90-i91, 2005. Brazzein is a high-intensity natural sweetener, having an excellent taste profile, 400-500 times the sweetness of sucrose, and excellent stability at low pH and high temperatures.

Described below are methods, using food-grade methodology and yeast fermentation, for enhanced production and improved purification of a natural, high- quality brazzein protein that enable the commercial production of brazzein. This unique protein has applications as a sweetener in the baking, beverage, and table-top product industries.

SUMMARY OF THE INVENTION

Disclosed is an expression vector for the production of a brazzein protein by a yeast cell comprising at least two cassettes, each cassette comprising a promoter sequence, a secretion signal sequence, a brazzein protein reading frame sequence, and a termination sequence, wherein the at least two cassettes are present in an expression vector comprising an integration sequence for integration of the cassettes into the yeast genome at a specific site.

Also disclosed is a recombinant P. pastoris cell comprising at least two copies of a gene construct comprising a pGAP promoter sequence, a secretion signal sequence, a brazzein Type II or Type III protein reading frame sequence, and a termination sequence, wherein the copies of the gene construct are present at a preselected integration site within the P. pastoris genome.

A method for producing enhanced levels of a brazzein protein is presented, comprising the steps of a) transforming one or more cells of P. pastoris with the expression vector of claim 1; b) growing the transformed cell or cells in a fermentation medium under aerobic conditions, wherein the transformed cells produce enhanced levels of a brazzein protein compared with cells containing only one copy of the brazzein protein gene.

A method for enhancing production of a foreign protein in recombinant Pichia is presented, comprising the steps of c) initiating a culture of the recombinant Pichia, under aerobic conditions, d) reducing oxygen levels in the culture to less than 5%, and e) maintaining oxygen levels in the culture at less than 5%. A method for purifying a protein from a yeast fermentation culture is also presented, comprising the ordered steps of a) adjusting the pH of the fermentation medium above or below the isoelectric point of the protein to be purified; b) removing the cells from the pH adjusted fermentation medium; c) subjecting the cell-free fermentation medium to ion-exchange chromatography and eluting the protein as a solution; d) passing the eluted protein solution through a first membrane to remove unwanted high molecular weight solutes; and e) concentrating and diafiltering the protein permeate on a second membrane to remove low molecular weight solutes.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 shows the construction of an expression vector for brazzein for integration into the His 4 site of the P. pastoris genome. Figure 2 is a schematic diagram of a process for purifying brazzein from fermentation medium.

Figure 3 is a graph showing the production of Type II brazzein by P. pastoris in 10 L reactor under oxygen stress. A. Raw data. B. Data normalized to optical density.

Figure 4 is a graph showing the production of Type III brazzein by P. pastoris in a 10 L reactor under oxygen stress. A. Raw data. B. Data normalized to optical density. Figure 5 is a graph showing the production of Type III brazzein by P. pastoris in a 1000 L reactor under oxygen stress. A. Raw data. B. Data normalized to optical density.

DETAILED DESCRIPTION OF THE INVENTION

Genetic enhancement of brazzein production

Genetic manipulation of P. pastoris uses a different set of techniques from those commonly used with bacteria. Foreign gene expression is engineered for stable integration by homologous recombination into specific sites in the host cell chromosome. These methods generally produce yeast cells that carry only one copy of the foreign gene. In some instances, foreign protein expression is enhanced by serendipitous integration of multiple gene copies in the yeast chromosome (Mansur, et al., Biotech. Lett. 27: 339-345, 2005; Sunga and Cregg, Gene 330: 39-47, 2004). However, this does not occur with all proteins and is unpredictable. Expression of integrated multiple 5 gene copies depends on the toxicity of the foreign protein for the yeast, the ability of the cell to process the pre-protein, and other factors, such as those reviewed in Inan, et al., Methods in Molec. Biol. 389: 65-75, 2007). Production of a number of foreign proteins by Pichia has been shown to reach a limit as the gene copy number increases, primarily because cellular mechanisms for protein secretion become saturated (Liu, et al., Appl.o Environ. Microbiol 71 : 5332-5340, 2005; Dean, et al., Genome Rsch. 11 : 1095-1099, 2001; Liu, et al., Protein Expression and Purification 30: 262-274, 2003). Therefore, it is not always possible to predict whether multi-copy integration will increase expression and secretion of a foreign protein in Pichia (Sreekrishna, et al., Gene 190: 55-62, 1997). A reference Pichia pastoris strain, PB3, contains one copy of an artificial genes based on the amino acid sequence of Type II brazzein and optimized for yeast codon usage. The brazzein gene is under the control of the pGAP (glyceraldehyde-3-phosphate dehydrogenase) promoter of Pichia pastoris, allowing constitutive expression of the brazzein gene, and downstream of a 19 amino acid pre-sequence for S. cerevisiae MF-α (mating factor alpha), allowing secretion of the brazzein into the yeast medium. o However, this yeast strain produces brazzein at insufficient levels for efficient commercial production.

Methods are described below that enhance the production of high levels of brazzein proteins by P. pastoris in large scale fermentation. The terms "enhanced production," "increased production," "enhanced yield," and "increased yield" as used ₅ herein mean that the amount of brazzein measured in the final Pichia fermentation broth by the "enhanced" or "increased" Pichia yeast production strain is greater than that of the reference strain, PB3.

One embodiment of a multi-gene construct that can be used to transform P. pastoris for enhanced production of brazzein is shown in Figure IA (Figure 1 depicts Typeo III brazzein, but the same strategy can be used for the Type II brazzein gene). Bam Hl gene cassettes are designed to contain, in order from 5' to 3', the P. pastoris pGAP promoter, the 19 amino acid S. cerevisiae mating factor alpha pre-sequence, either the type II or type III brazzein gene and the AOXl terminator, as shown in Figure IA. A small percentage of Type II brazzein is spontaneously converted to Type I brazzein in ₅ yeast fermentation cultures. However, Type I brazzein can be produced by chemically converting the N-terminal glutamine of Type II brazzein to pyroglutamate using standard methods, for example, as described in Khandke, et al., Int. J. Peptide Protein Res. 34: 118-123, 1989 or through enzymatic conversion, for example, with the Qcyclase™ enzyme from Qiagen. While the pGAP promoter is preferred, other P. pastoris constitutive or inducible promoter sequences may be used, including, but not limited to, PPHIS4, FLDl, PEX 8, YPTI, as described in Sunga and Cregg, Gene 330: 39-47, 2004 and Lin-Cereghino, J and Lin-Cereghino, GP, Methods in Molecular Biology 389: 11-25, 2007, TEFl (Ahn, et al., Appl. Microbiol. Biotechnol. 74: 601-608, 2007), and PGKl (de Almeida, et al., Yeast 22: 725-737, 2005; WO/1994/24280). More than one type of promoter may be used in multi-copy constructs, e.g., the pGAP promoter with PGKl. Alternatively, hybrid promoters may be created using parts of different promoters.

Any leader signal sequence (pre-, pro-, or pre-pro), from Pichia or other organisms, that promotes secretion of brazzein can be used, such as the full length S. cerevisiae MF-α or other variations of the MF-α leader sequence; matrix metalloproteinase, PHO-I and PHA-E leader sequences; and the β-glucanase leader sequence. (Xiong et al., Acta Biochim. Biophys. Sinica 35: 154-160, 2003; Sreekrishna, et al., Gene 190: 55-62, 1997; Lin-Cereghino, J and Lin-Cereghino, GP, Methods in Molecular Biology 389: 11-25, 2007; US2007/0122895).

In addition to AOXl, other terminator sequences can be utilized. For an appropriate promoter, such as those listed above, the termination signal for the associated P. pastoris gene can be used. For example, the FLDl promoter and terminator have been successfully used together as described by Sunga and Cregg, Gene 330: 39-47, 2004. Alternatively, terminator and promoter sequences from different genes may be paired in the construct.

Multimeric brazzein cassettes are prepared by ligating copies of the Bam Hl fragments to each other. Ligations may be replicated, screened and characterized by any appropriate method, such as those described in Example 1. Following construction, the cassettes are then cloned into an expression plasmid or other vector from which the DNA can be easily amplified. Any appropriate vector carrying a selectable marker, e.g., antibiotic resistance, may be used. For example, the cassettes can be cloned into E. coli plasmid vectors and E. coli used to produce the DNA for transforming yeast cells. In the embodiment shown in Figure IB, ligation products with the desired number of brazzein cassettes were inserted into the Bam Hl site of pUC57, as described in Example 1. Alternatively, the DNA can be amplified in vitro, for example, as described by Dean, Genome Res. 11 : 1095-1099, 2001, and used to transform yeast cells.

In one embodiment, an expression plasmid is created by inserting the Pichia histidine gene 4, PPHIS4, which includes the HIS4 reading frame, 84 base pairs of promoter sequence and 40 base pairs of terminator sequence, into the vector comprising the brazzein cassettes, (Figure 1C). PPHIS4 allows integration of the foreign gene at the chromosomal HIS 4 site of P. pastoris. Any P. pastoris gene containing appropriate restriction sites can be used as a homologous recombination site for integration. Other possible integration sites include, but are not limited to, P. pastoris genes pGAP, AOXl, PHOl, and FLDl. The final expression plasmid is generally sequenced to confirm that the correct gene sequences are present.

Conventional methods of molecular biology are used to create the brazzein cassettes, plasmids, and expression vectors. Such methods are described, for example, in Molecular cloning : a laboratory manual, 3rd ed., Sambrook, Joseph; Russell, David W., New York: Cold Spring Harbor Laboratory, 2001; Molecular Cloning: A laboratory manual, 2d ed., Sambrook, J., Fritsch, E. F., and Maniatis, T. New York: Cold Spring Harbor Laboratory Press, 1989; Cold Spring Harbor Protocols, cshprotocols.cshlp.org; Current Protocols in Molecular Biology, currentprotocols.com. The expression vectors are used to transform P. pastoris. Any appropriate method of transformation may be used including electroporation and chemical-mediated transformation as described, for example, in Cregg, Methods in Molecular Biology 389: 27-42, 2007 and in Becker and Guarenta, Methods in Enzymology 194: 182-187, 1991. Transformed yeast cells may be stored at -80° C in a solution containing 10% glycerol.

Environmental enhancement of brazzein production

The production of brazzein by Pichia yeast engineered to produce brazzein proteins under control of the pGAP promoter can, surprisingly, be regulated by controlling the availability of oxygen in the fermentation reactor. In general, fermentation cultures of Pichia are grown as fed batch cultures in 20-40% air saturation (0.04-0.09 mg/ml dissolved oxygen). As described in Examples 5 and 6 and shown in Figures 2-4, restricting available O ₂ to less than about 5% air saturation causes a 2 - 2.5 fold increase in production and secretion of brazzein Type II (Figure 2) and Type III (Figures 3 and 4).

No differences were observed in optical density {i.e., cell proliferation), CO ₂ evolution, O ₂ uptake, or glucose consumption between the control and oxygen stressed cultures. The yeast continue to grow aerobically in the low O ₂ environment, and no changes in cell metabolism other than enhanced production of brazzein were observed. Although the exact mechanism for the oxygen stress effect is not known, it appears to have a specific effect on the synthesis of the foreign proteins. Secretion of the foreign proteins occurs at a maximal rate even at normal oxygen levels, as evidenced by the lack of accumulation of these proteins inside the yeast cells. While not excluding other hypotheses, it appears that oxygen stress affects the function of the pGAP promoter in Pichia. Therefore, the oxygen stress technique can be used to increase Pichia production of any foreign protein that is controlled by the pGAP promoter. Other promoters are likely to be stimulated by oxygen stress.

Pichia is one of the best organisms for large-scale production of foreign proteins, and genetic methods to enhance the production of foreign proteins in Pichia have been widely studied and are reviewed in Cregg, Methods in MoI. Biol. 389: 1-7, 2007. The pGAP promoter is a constitutive promoter native to Pichia. Standard methods of molecular biology, such as those described above and in Examples 1 - 4, may be used to clone a foreign protein with the pGAP promoter in an appropriate expression vector for production and secretion of the foreign protein under control of the pGAP promoter. Pichia cells are then transformed with the expression vector using any suitable method of transformation and grown in a fed-batch culture, such as that described in Example 5. Other methods of culture that avoid the production of ethanol may also be used, for example, a simple batch culture with glycerol as the food source, or a chemostat culture system. The yeast culture is usually maintained and fed until an upper limit of cell density corresponding to a wet (packed) cell volume that is 60% of the total fermentation volume is achieved.

Oxygen stress of the culture can be initiated at any time. Oxygen stress can be initiated by any means that maintains O ₂ availability in the culture at less than 5%. In one embodiment, described in Example 5, oxygen stress was initiated during the last feed and was achieved by controlling the agitation speed of the culture. Oxygen availability can also be reduced by limiting the flow of air, manipulating the composition of the gas, altering the backpressure on the reactor, or, to some extent, by altering the feed rate of glucose.

Enhanced purification of secreted proteins from large-scale fermentation broth

High yields of purified protein can be obtained from large-scale yeast fermentations by following a series of ordered steps, which are shown schematically in Figure 2. The process may be scaled to any size, so it can be used for any fermentation volume, and is suitable for any protein. Pichia is a preferred organism for production and secretion of foreign proteins into fermentation medium. Pichia will grow in a minimal salt medium, allowing filtration and chromatography at low conductivity. In addition, Pichia does not naturally secrete small proteins into the medium. Therefore, the only protein present in the medium in significant amounts will be the foreign protein to be purified and purification procedures can be optimized for that particular protein. One advantage of the process is that it does not utilize harsh chemical steps and can be employed in food production.

The first step of the process is performed before the yeast cells are removed from the broth. The pH of the fermentation medium is adjusted to create either a positive or negative charge on the protein to be purified. Thus, the pH of the medium will be below or above the isoelectric point (pi) of the protein for a positive or negative charge, respectively. Adjusting the pH as a first step reduces the need to dilute the micropermeate, thereby reducing the volume of material to be processed later through the ultrafiltration membrane. The second step is to remove the cells from the medium. This may be accomplished by any appropriate method including centrifugation, clarification, and microfiltration through filters with various pore sizes. For microfiltration processes, flux rates can be regulated to match the rate of flow through the chromatography column directly downstream of the membrane.

The permeate generated by microfiltration is pumped directly to a chromatography column. Any size column can be used and size selection is based on parameters known in the art, such as volume of permeate, capacity of column resin, rate, number of load-unload cycles desired, etc. Flow rate can be regulated by a feed pump that feeds the column, thereby also limiting the permeate flow rate from the microfiltration membrane. Any cation or anion exchange resin with a high capacity for the protein of interest at the pH of the permeate can be used. As the permeate runs through the column, the protein to be purified binds to the exchange resin while nucleic acids and polymers, carbohydrates, and most other proteins that may be in the cell broth pass through the column. When a sufficient volume of permeate has been generated, the feed of permeate to the chromatography column is stopped and the column is washed with several volumes of low salt buffer at a pH approximately that of the permeate to remove residual permeate from the column without removing bound protein. The column is then washed with several volumes of salt buffer having an intermediate molarity, e.g. , 0.25-0.35M NaCI in 0.05 M acetate, at a pH approximately that of the permeate to remove bound proteins that are not the protein to be purified. After the washing steps, the column is eluted with a high salt buffer at a pH approximately that of the permeate to remove the bound protein to be purified. In lieu of a high salt elution buffer, the pH of the buffer can be raised or lowered to reverse the charge on the bound protein and release it from the resin. Wash and elution buffers and other column chromatography parameters can be optimized for specific proteins by methods known in the art, e.g., as described in Ion Exchange Chromatography & Chromatofocusing: Principles and Methods, GE Healthcare, www.4. gelifesciences.com/aptrix/upp00919.nsf/Content/71857706466Dl AB8C1256EB40 041805D/$file/11000421AA.pdf (2004). The chromatography process removes most of the waste compounds generated during fermentation. This allows the subsequent ultrafiltration steps to run faster and minimizes the passage of contaminants into the ultrafiltration permeate.

The eluate is collected in fractions until all of the bound protein has been eluted. The process may be varied by using multiple columns, a simulated moving bed, or any appropriate modification. Fractions are assayed for protein according to known methods, e.g., absorbance at 280 nm, and fractions containing protein can be pooled for further purification. The pH of the protein-containing fractions may optionally be adjusted to a pH close to the pi of the protein to facilitate ultrafiltration at this step.

The protein fractions are next passed through a 5 to 30 kD nominal molecular weight cut-off (NMWCO) ultrafiltration membrane with a high volume of reverse osmosis purified (RO) water to remove co-eluted Pichia proteins. Any type of ultrafiltration membrane may be used. Such diafiltration methods and methods for optimizing diafiltration parameters are known in the art, e.g., as described in Technical Report PN33289, Pall Life Sciences, pall.com/pdf/02.0629_Buffer_Exchange_STR.pdf). Following this first ultrafiltration step, the permeate is concentrated and diafiltered with a low NMWCO membrane, in general, IkD or less NMWCO. The protein is retained and salts and other low molecular weight contaminants pass through the membrane. The concentrated protein solution can then be diafiltered with water or buffers, if desired, to adjust the salt concentration in the concentrated solution. It may be desirable to again adjust the pH of the permeate prior to concentration to maintain high flux through the membrane and to prevent protein precipitation in the concentrated solution, as described in Example 9.

If desired, the concentrated solution of purified protein may be dried by any appropriate means such as oven drying, spray drying, and freeze drying for storage as a dry product.

In one embodiment of the process, the protein to be purified is brazzein. An exemplary process for purifying types I, II, and III brazzein is described in detail in Examples 7-9. For types I, II, and III brazzein, the pH of the fermentation medium is adjusted to 3.7-3.9 prior to removal of the yeast cells. A sample of the broth is assayed to estimate the concentration of protein in the broth. The fermentation broth is then pumped through a 2 m ² microfiltration membrane (0.22 μm), at a rate appropriate for the estimated protein concentration, to remove cells. The permeate is then directly fed to a cation-exchange chromatography column. After completion of permeate input, the column is washed with a no salt and then low salt acetate buffer, pH about 3.8. Brazzein is eluted from the column with a high salt (IM NaCI), acetate (0.05 M) buffer, pH about 3.8. Following the chromatography step, the brazzein proteins are about 95% pure. In this embodiment, the pH of the brazzein-containing fractions is adjusted to approximately 6.0-6.2 prior to concentration and ultrafiltration. A 1OkD NMWCO PES (polyethersulfone) filter is used to concentrate the volume of the brazzein-containing fractions. PES filters are preferred, because they do not exhibit changes in porosity, are easily cleaned and can be heat-sterilized. The solution is diafiltered to remove salts and buffer ions and the permeate is then concentrated and diafiltered a second time to flush the remainder of the brazzein through the membrane. The permeate is then concentrated with a 1 kD NMWCO PES filter which is then flushed with purified water. For Type III brazzein, the permeate from the 1OkD filter is adjusted to pH about 4.0 before concentration and filtration through the 1 kD filter membrane to prevent protein precipitation in the final concentrate. A fermentation volume of 620 L containing a mixture of Types I and II brazzein (about 280 g) yielded 4 L of solution containing about 260 g brazzein at 98% purity at the end of the purification process (Examples 7 and 8). For Type III brazzein, a fermentation volume of 600 L containing about 126 g of brazzein yielded 2 L containing about 99 g of Type III brazzein at 98% purity (Example 9). The final solution can be dried or filter-sterilized and stored as a solution.

Examples 1. Construction of a multi-cassette expression vector for enhanced expression of brazzein

Bam Hl brazzein gene cassettes were designed to comprise, in order from 5' to 3', the P. pastoris pGAP promoter (SEQ ID NO: 1), the 19 amino acid S. cerevisiae alpha mating factor pre-sequence (SEQ ID NO:2), either the type II (SEQ ID IMO:3 (includes stop codon)) or type III (SEQ ID N0:4) brazzein gene and the AOXl terminator (SEQ ID NO: 5), as shown in Figure IA. The brazzein cassettes were chemically synthesized and cloned into pUC57 by Integrated DNA Technologies, Coralville, IA.

Multimeric brazzein cassettes were prepared by excising the Bam Hl fragment by digestion with Bam Hl, gel purifying the fragment, and ligating copies of the Bam Hl fragment to each other. The ligation reaction was purified and cloned into Bamhl linearized, dephosphorylated pUC57 using standard techniques of molecular biology. The resulting pUC57 vector (Figure IB) was used to transform stbl4 E. coli cells (Invitrogen). Transformants were identified on X-gal/IPTG spread LB+ampicillin plates by blue/white color selection. Positive colonies were confirmed by restriction digests of miniprep cultures. Because EcoRI and Sail cut to the outside of the βamHI sites, and

EcoRl/Sa/I digest can be used to determine the number of brazzein copies present. BgIU cuts near the 3' end of brazzein and a BgIlI digest can be used to determine the orientation of the cassettes relative to each other. Results suggest that only head-to-tail repeats are tolerated by the Stbl4 cells. The brazzein cassettes were sequenced to verify that the correct sequences were present.

To produce an expression plasmid, the Pichia histidine gene 4, PPHIS4, (SEQ ID NO: 6), which includes the HIS4 reading frame, 84 base pairs of promoter sequence and 40 base pairs of terminator sequence was cloned into the pUC57/Brazzein cassette construct (Figure 1C). An expression plasmid was produced for Type II brazzein with 4 copies of the brazzein Type II cassette. A Type III brazzein plasmid was produced with 2 copies of brazzein Type III cassette. Expression plasmids with greater numbers of copies of either the Type II or Type III brazzein gene could be made by ligating additional copies of the Bam Hl cassette together and inserting them collectively into a vector for yeast transformation. 2. Transformation of P. pastoris with brazzein expression plasmids

P. pastoris cells were transformed with the Type II or Type III brazzein expression plasmids of Example 1 via electroporation. Plasmids were cut with Nhel to target insertion to the HIS4 site. Aliquots (40 μl) of competent P. pastoris cells suspended ice- cold IM sorbitol were incubated with 10 μg of linearized plasmid DNA in an ice-cold 0.2 cm electroporation cuvette for 5 min on ice. Electroporation was performed using the following parameters: 25 μF, 200 Ω, 2000 v. Immediately following electroporation, 1 ml_ of ice-cold IM sorbitol was added and the contents transferred to a sterile 10 ml_ tube. The tube was incubated at 30° C for 2 h then aliquots were spread onto BMDS plates (100 mM potassium phosphate, pH 6.0, 1.34% yeast nitrogen base without amino acids, 0.5% ammonium sulfate, 4 x 10 ^"5 % biotin, IM sorbitol, 2% dextrose, 2% agar). Plates were incubated for 4-5 days at 30° C.

In some instances, chemical transformation of Pichia was carried out using a commercial yeast transformation kit (EZ Yeast Transformation Kit, Zymo Research) according to manufacturer's instructions.

3. Identification of transformed P. pastoris

Yeast colonies from Example 2 were grown for 48 h in 10 mL of YPD (yeast extract, peptone, dextrose) medium and transformants were identified by dot-blotting. Culture medium was centrifuged to remove cells and cell-free aliquots (2 μl) of full- strength and diluted (1: 10) medium from each colony were blotted adjacent to each other onto a nitrocellulose membrane using a dot-blot filter apparatus. The membrane was blocked for 30 min with "blocking solution," (Western Breeze kit, Invitrogen). The membrane was then washed twice with water and incubated for 1 h in a 1 : 100,000 dilution of primary antibody (rabbit-anti-brazzein) capable of detecting Types I-III of brazzein in blocking medium. Following incubation with primary antibody, the membrane was then washed three times with "washing solution," (Western Breeze kit, Invitrogen), and incubated with a 1 : 10,000 dilution of secondary antibody (goat-anti- rabbit-HRP, Jackson Labs) for 30 min at room temperature. The membrane was subsequently washed three times with the washing solution and twice with water. A chromogenic substrate solution, (Western Breeze kit, Invitrogen), was applied to the membrane and incubated until a purple color developed (1-60 min). The membrane was again washed twice in water, air dried, and examined visually. Medium from colonies of yeast carrying multiple copies of the brazzein genes contained a noticeably higher amount of brazzein compared with the reference strain grown under identical conditions. These results were confirmed and quantified by gel electrophoresis and HPLC.

HPLC results for high-producing colonies are shown in Table 1. 4. Production of brazzein proteins by recombinant P. pastoris

The transformed yeast of Example 3 were cultured in 1 L stir-tank reactors using a fed batch system with medium comprising basal salts and dextrose for about 60 h. Fermentation medium was analyzed for brazzein content by polyacrylamide gel electrophoresis and HPLC analysis. Cells were removed from the broth samples by centrifugation (3,000 x g, 10 min, room temperature). The supernatant was diluted, filtered, and applied to a cationic HPLC column. The integrated peak eluting with an authentic brazzein sample was converted to mg/L using a standard of pure brazzein of known concentration. Results from the HPLC analysis are shown in Table 1. "Fold increase" is increase relative to the PB3 reference strain grown under the same conditions. The data show that the yeast with four copies of the Type II brazzein construct each secreted over four-fold greater amounts of Type II brazzein into the fermentation medium than the reference strain, PB3.

Table 1

Amount of Type II brazzein secreted into fermentation medium by recombinant P. pastoris

5. Enhanced production of brazzein by recombinant P. pastoris

Recombinant P. pastoris capable of expressing and secreting brazzein proteins were created as described in examples 1-4, and grown as high density, fed batch fermentation cultures. The recombinant yeast were added to a minimal yeast fermentation medium comprising glucose as the sole food source. Two, 10 L fermentation reactors were prepared by adding 1 L of 1OX salts buffer containing 120g/L of phosphoric acid and 25 g/L of K ₂ SO ₄ , to 6.5 L of reverse osmosis purified (RO) water. The reactor and contents were sterilized at 121°C for 30 min. A sterile solution containing 2050 mL of 72% dissolved solids as 95% dextrose, 250 mL of a 500g/L MgSO ₄ (7H2O) solution, and 200 mL of trace metals with biotin was prepared. This solution (25 mL) was added to each fermentation tank, and the pH was adjusted to pH 5.6 with 28% NH ₄ OH. Each reactor was inoculated with 100 mL of an overnight culture (O. D. ~ 5) of the recombinant Pichia cells. The yeast cells were allowed to grow overnight (about 16h) until glucose in the medium was completely consumed. The medium was maintained at pH 5.6 and 30% air saturation (regulated by stirring).

The dextrose feed solution was pumped into the reactor at increasing rates (adjusted every 4 hours) until about 1 L had been added to each reactor (about 48 h). A sample was then taken from each reactor and brazzein levels in the medium were measured. The sample was filtered through a 0.22 μm filter and assayed on HPLC by ion exchange chromatography. The feed rate of the culture was increased to a final value of about 100 ml/h. This rate was continued until the culture was terminated (about 16h). In the "control" reactor the oxygen level was maintained at 30% by regulating the stirring rate. Table 2 lists the approximate feed and agitation rates over time for a 6Oh, 10 L, control culture. In the "oxygen stress" reactor the agitation speed was locked at 750 RPM after reaching that rate of speed (Steps 7-8 of Table 2). In the oxygen stress reactor, oxygen levels dropped to as low as 3% during the last phase of the culture. Respiratory quotient was elevated about 5% under oxygen stress. Only minor differences were observed in optical density, CO ₂ evolution, O ₂ uptake, or consumption of base between the control and oxygen stress reactors.

Table 2

Simulated exponential fed batch culture for 1OL final volume process

EFT - elapsed fermentation time

After all the glucose feed (2500 mL) had been added, samples of the medium were again taken and analyzed as described above to quantify brazzein in the cell-free broth. Results for cultures of Pichia expressing Type II brazzein are shown in Figure 3 and for Type III brazzein in Figures 4 and 5. The amounts of Type II or Type III brazzein did not differ between the control and oxygen stressed reactors at 48h, at which time conditions in each reactor were about the same. However, at the end of the culture (about 6Oh), the amount of Type II brazzein in the oxygen stress reactor (~ 3% O ₂ ) was approximately twice the amount of Type II brazzein in the control reactor (~30% O ₂ ) as

5 shown in Figure 3A. On a per cell basis, Pichia cells produced about 2.2 times more Type II brazzein under oxygen stress conditions (Figure 3B).

Similar results were obtained for Pichia producing Type III brazzein. Oxygen stressed reactors contained approximately twice the amount of Type III brazzein as the control reactors as shown in Figure 4A. On a per cell basis, Pichia cells produced abouto 2.1 times more Type III brazzein under oxygen stress conditions (Figure 4B).

6. Effect of oxygen stress on brazzein production in commercial scale fermentation

The experiment described in the previous example was performed in a 1000 L fermentation reactor, i.e., a 100-fold scale-up of the process, with Pichia engineered to produce Type III brazzein. The reactors were maintained at 30% O ₂ until 100 L ofs glucose had been added (agitation rate of about 250 RPM). At this time, the agitation rate was regulated to maintain O ₂ at about 3% for the remainder of the culture. As shown in Figure 5A, approximately 2.5 times more Type III brazzein was produced under oxygen stress conditions. Figure 5B shows that this effect is due to increased Pichia productivity, as per cell production of Type III brazzein also increased about 2.5 fold. o 7. Initial purification of Type I and Type II brazzein from fermentation medium

A 1000 L fermentation of P. pastoris carrying 4 copies of the type II brazzein gene was carried out as described in Example 5, but with MgSO ₄ reduced by a factor of 5, at 30% dissolved oxygen. The pH of the medium was allowed to drop and was then maintained at 3.7 -3.9 by addition of 12 IM HCI. A sample of medium was filtered5 through a 0.22 μm filter and assayed by ion exchange HPLC. Peak areas corresponding to both Type II and Type I brazzein were found, as about 25% of Type II brazzein is, in general, spontaneously converted to Type I brazzein during culture. The titer of total brazzein protein was estimated to be about 0.45 g/L of cell free broth. This estimate extrapolates to an estimated 28Og total brazzein in the fermentation. o Whole cell broth was pumped out of the reactor into a 2 m ² microfiltration membrane (0.22 μm, Millipore Duropore membrane) at a feed rate of about 45 LPM with a centrifugal pump. The feed pressure was about 37 psig and the retentate pressure was about 16 psig. Retentate was recycled back to the reactor. The permeate was pumped to a Waters Quickscale column containing 5 L Toyopearl SP550C (Tosoh Bioscience) ₅ cation-exchange chromatography column at 1 LPM. After about 8h the permeate input to the column was stopped and the column was first washed with four volumes (20 L) of 0.005M acetate buffer, pH 3.8 (+/- 0.1), and then washed with 20 L of 0.25 M NaCI in 0.05 M acetate buffer, pH 3.8. After washing, the column was eluted with 20 L of 1 M NaCI in 0.05 M acetate buffer, pH 3.8. Eluate was collected in 1 L fractions and each was assayed for brazzein. The column was flushed with eluting buffer and then washed with RO water. Significant amounts of brazzein were found only in the first 11 liters of eluate, and peak levels were found at 10 liters. The results of the assays are shown in Table 3. A total of 160-17Og of brazzein was collected. Type I brazzein eluted early and later fractions contained primarily Type II brazzein.

10 Table 3 First elution of brazzein from ion-exchange column

The permeate line was reconnected to the column and the process was continued for another 12 h. The column was washed and eluted as described above and an

I ₅ additional 8 L of eluate containing brazzein were collected. An additional 110-12Og brazzein was collected (Table 4). The brazzein collected from the column in both elutions was about 95% pure as determined by absorbance at 280 nm.

Table 4

20 Second elution of brazzein from ion-exchange column

8. Further purification of Type I and Type II brazzein

The purified brazzein from Example 7 was subjected to a two-step filtration process to remove salts and buffer ions. First, fractions were pooled to yield batches containing approximately 60-8Og of brazzein. The pH of each batch was adjusted with 1 N NaOH to 6.0-6.2. A 10 kD PES filter, operating at 30-40 psid, was then used to concentrate the volume of solution to about 2.5 L. The solution was then diafiltered with 20 L of RO water and the permeate collected. The permeate was again concentrated and diafiltered to flush the remainder of the brazzein through the membrane and into the permeate. Little or no brazzein was found in the retentate.

The permeate was concentrated to 500 mL with a 1 kD PES filter cartridge and the system was flushed with 500 mL of RO water. Samples were assayed for brazzein. In general, less than 5% of the brazzein was lost to the 1 kD permeate and 95% was recovered in the retentate. The complete purification process and results are summarized in Table 5. A final yield of approximately 260 g of brazzein was recovered at 98% purity.

Table 5 Type I and II Brazzein yields through purification process

9. Purification of Type III brazzein from recombinant P. pastoris

The purification process described in Examples 7 and 8 was carried out with 1000 L fermentation medium from a culture of P. pastoris carrying two copies of the gene for Type III brazzein. At the end of the fermentation, the pH of the medium was adjusted to about 3.8. An assay of a filtered sample of the medium indicated that the reactor contained about 126 g of brazzein. The medium was passed through a 2 m ² microfiltration cartridge as described above, but at a permeate rate of about 1 LPM. The flow rate of permeate was maintained for about 22 h.

The permeate was passed directly through a 5 L column of SP 550C resin. The column was then washed and eluted as described in Example 7, but the NaCI concentration in the second wash buffer was 0.35 M instead of 0.25 M. Eluate was collected in 1 L bottles and assayed by A280 for total protein and by HPLC for brazzein.

Elution samples were adjusted to pH 6.1 (+/-0.1) with 1 N NaOH. Two groups of 6 bottles each were processed through a 10 kD NMWCO PES filter and diafiltered with about 40 L of RO water in two steps to produce a permeate containing substantially all of the brazzein. This permeate was collected from the membrane, adjusted to pH 4.0 (+/- 0.2) using 1 N HCI, concentrated to 1 L using a 1 kD membrane, and subsequently diafiltered with about 10 L RO water. Diafiltration was continued until conductivity reached about 250 μS/cm or less. The solution was then allowed to concentrate to about 500 ml. and the system was flushed with 500 ml_ of RO water. The liquid product was freeze dried to below 5% moisture. The results of this purification are shown in Table 6. Final yield was 2 L containing approximately 99 g of 98% pure Type III brazzein.

Table 6

10 Type III Brazzein yields through purification process

Although the invention is illustrated and described herein with reference to specific embodiments, the invention is not intended to be limited to the details shown.

I ₅ Rather, various modifications may be made in the details within the scope and range of equivalents of the claims and without departing from the invention.