Background The present invention relates in general to methods for extraction of Hepatitis B surface Antigen (HBsAg) and in particular to methods for extracting HBsAg which employ nonionic detergents.

The hepatitis B virus causes a disease now known as hepatitis B, but formerly known as "serum hepatitis". It has been estimated that there are more . than 200,000,000 people who persistently have hepatitis B virus in their blood. Infection with the virus is a major cause of acute liver disease. Carriers of hepatitis B virus have a high risk of contracting cirrhosis and hepatocellular carcinoma.

The human hepatitis B virus has been identified with the "Dane" particle which is found in the serum of carriers and which is the causative agent of clinical hepatitis B infection.

The Dane particle is a 42 nanometer membrane structure which includes lipids, DNA, and at least four proteins: hepatitis B surface antigen (HBsAg), hepatitis B core antigen (HBcAg), hepatitis B e antigen (HBeAg), and a DNA polymerase. In their serum, carriers also have 22 nanometer lipid particles which contain HBsAg but not DNA, HBcAg, HBeAg or DNA polymerase. Hepatitis B vaccines currently in use employ 22 nanometer particles obtained from human plasma.

Human plasma used in the manufacture of hepatitis B vaccine has an antigen concentration of

about 400 micrograms per milliliter. Because the total serum concentration is about 60 milligrams per liter, only a 150-fold purification is required. Wampler, et al., in "Modern Approaches to Vaccines," Chanock, et 5 al., eds.. Cold Spring Harbor Laboratory, Cold Spring

Harbor, New York, pp. 251-256 (1984). However, inasmuch as the starting material is human plasma, plasma-derived hepatitis B vaccines are limited in supply and extreme caution must be exercised to insure that they are free 10 of all harmful contaminating material, including infectious viruses.

Another approach to obtaining hepatitis B vaccine involves infecting cultured malignant cells, specifically hepatoma cells, with hepatitis B virus.

15 HBsAg is harvested from preparations of lysed hepatoma cells. Knowles, et al., PCT Application No. PCT/US81/00778. Although this approach eliminates, the need for human plasma, the undesirability of using cancer cell products in vaccines, the need for extreme

20 caution to avoid infectivity and the difficulties inherent in mammalian cell culture limit the usefulness of this approach.

Monkey kidney cells transfected with - _c recombinant plasmids containing the gene for HBsAg liberate HBsAg by lysis or by secretion. Levinson, et al., European Patent Application No. 73,656. Nevertheless, with the exception of the elimination of the concern over the use of products of cancerous cells _ ₀ in vaccines, production of hepatitis B vaccine from monkey kidney fibroblasts shares the drawbacks inherent in the production of hepatitis B vaccine from hepatoma cells.

In light of the problems with conventional 35 vaccine production, it is desirable to have methods for preparation of HBsAg in isolation from other components

of the hepatitis B virus. Production of HBsAg in microorganisms by the use of recombinant technology permits such isolated production.

In one approach to applying recombinant technology to HBsAg production, the gene coding for HBsAg may be inserted into a bacterial plas id and may be amplified and expressed in several Escherichia coli host organisms. Rutter, et al., European Patent Application No. 020,251. However, this technique results in low yields because HBsAg is easily degraded within E. coli and the growth of E. coli is inhibited by HBsAg. Miyanohara, et al., European Patent Application No. 105,049. In addition, certain bacterial cell components, for example lipopolysaccharides, are highly toxic to humans and pose purification problems. Moreover, bacteria, being prokaryotes (i.e., members -of the group of organisms which lack in nucleus) may provide inefficient translation of the genes of eukaryotes (i.e., members of the group of organisms which possess a nucleus) inasmuch as: bacteria cannot perform certain processes, such as splicing out of introns or proteolytic clevage of precursor proteins; bacteria do not glycosylate, phosphorylate or methylate proteins, all of which are post-translational modifications which may be important for the immunogenicity of proteins; bacteria do not recognize the so-called signal peptide which is important for secretion of gene products in eukaryotes; and codon preference (i.e. the facility with which a particular sequence of nucleic acids coding for an amino acid constituent of a protein is expressed) may be different in prokaryotic and eukaryotic organisms. HofSchneider, et al., European Patent Application No. 105,141. Another approach to expression of HBsAg in microorganisms, and one which overcomes almost all of

the objections to the use of E. coli as a host, employs a yeast expression system. Such expression systems have generally involved the use of bacterial-yeast shuttle vectors, which are plasmids having sequences permitting replication in both bacteria and yeast. Rutter, et al., European Patent Application No. 072,318; Hitzenman, et al., European Patent Application No. 073,657; Cabezon, et al., European Patent Application No. 106,828; and Bitter, et al., Gene, 3_2, 263-274 (1984). Yeast offers several advantages for the production of eukaryotic gene products: yeast is readily grown in culture in large quantities; the technology of yeast culture on a large scale is well understood; because yeast cells are eukaryotic, they contain processing machinery for glycosylation, phosphorylation and methylation; and yeast cells better tolerate the HBsAg protein. Rutter, et al., supra; Hitzeman, et al., supra Furthermore, despite concerns that yeast-produced HBsAg might _' have a lower immunogenic potential than the form produced in mammalian cells (HofSchneider, et al., supra) , yeast-derived particles have an ED _5Q (a measure of the effective dose necessary to elicit an immune response) of 112 ng which is only slightly higher than the ED _5Q of 98 ng for mammalian cell-derived particles. Burnette, et al., in "Modern Approaches to Vaccines", Chanock, et al., eds., Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, 245-250 (1984).

Protein micelles are aggregations of protein molecules having hydrophobic portions, which avoid contact with water molecules, and hydrophilic portions, which readily associate with water molecules. In a solution, the hydrophobic portions of the proteins in micelles become oriented toward the center of the micelle while the hydrophilic portions of the proteins are oriented toward the surface of the micelle.

Micelles formed from HBsAg protein molecules produce consistently higher levels of antibodies than are induced by intact Hepatitis B particles. Skelly, et al., Nature, 290, 51-54 (1981). A micellar vaccine prepared from yeast HBsAg micelles has a slightly lower EDJ _Q (22 ng) than micelles prepared from mammalian cells (25 ng). Burnette, et al., supra.

The process of extraction of polypeptides from the host cell is of importance for the preparation of polypeptide vaccines in general, and for the formation of micelles in particular. Detergents may be employed for the extraction of membrane proteins. Different types of detergents have different effects upon the nature and activity of protein aggregates formed. Three detergents which may be employed for protein extraction are bile salts, ionic detergents and nonionic detergents.

Membrane proteins solubilized by bile salts may be precipitated out of salt solutions over a wide range of salt concentrations, a property useful for the separation of protein fractions. Tzagoloff et al., Methods in Enzymology, 22, 219-230 (1971). However, bile salts may dissociate integral membrane protein complexes so that native properties of an extracted protein, such as the state of aggregation, may not be maintained. Helenius et al.. Methods in Enzymology, 56, 734-749 (1979).

Ionic detergents are very effective solubilizing agents for membrane proteins. Tzagoloff et al., supra. However, ionic detergents nearly always denature proteins at the concentrations and temperatures required for complete solubilization of membranes. Helenius et al., supra.

Nonionic detergents are effective in dissociating lipids from proteins, but are relatively

ineffective at disrupting protein-protein interactions. Helenius et al., supra. Thus, nonionic detergents are especially useful in preserving the structure of protein micelles during the extraction process.

A nonionic detergent which may be employed to isolate viral membrane proteins is sold under the name Triton X-100 ^® , a trademark of Rohm and Haas Co., Philadelphia, Pennsylvania. Triton X-100 ^® is an example of a group of commercially available, nonionic detergents having polar groups containing pαlyoxyethylene. Members of this group are almost always heterogenous in composition in that they contain a wide distribution of polyoxyethylene chain lengths. The chain length for a member of the group is commonly specified as an average chain length. For example, Triton X-100 ^® may be -generically described as a polyethylene glycol p-isooctylphenyl ether having an average polyoxyethylene chain length of 9.6. Helenius et al., supra, contains lists of polyoxyethylene- containing detergents and information regarding their properties. Helenius et al, supra, is incorporated by reference herein.

Triton X-100 ^® is used at various concentrations for the extraction of membrane proteins: at 0.05%, 0.3% and 2% for Semliki forest virus membrane proteins [Simons et al., J. Mol. Biol., 80,119-133 (1973); at 5% (v/v) for influenza virus envelope protein [Larin et al., J. Hy . Camb. , 69_, 35-46 (1971)]; at 2% with EDTA [Schnaitman, J. Bacteriol. ,108, 553-563 (1971)] or without EDTA [Schnaitman, J. Bacteriol., 108, 545-552 (1971)] for E. coli cytoplasmic membrane proteins and for E. coli cell wall proteins; at 0.02%, 0.03%, 0.04% and 0.05% for tick-borne encephalitis envelope proteins [Heinz et al., Acta virol. , 23, 189-

197 (1979)]; at 0.05% and 20% with EGTA for filament proteins in Ehrlich ascites tumor cells [Traub et al., J. Cell Sci., 53, 49-76 (1982)]; and at 0.1% (v/v) for Sendai virus F protein [Welling et al. , J. Chromatog. , 266, , 629-632 (1983). Although this list is not exhaustive, it does indicate that it is not a straightforward matter to determine the appropriate concentration of and conditions for use of a nonionic detergent in the extraction of a protein. This is particularly true for certain types of membrane-bound proteins, such as nucleocapsid proteins, Helenius et al., Biochim. Biophys. Acta, 307, 287-300(1973).

Following work on micelle preparation in Semliki forest virus [Helenius et al., Biochim. Biophys. Acta, 307, 287-300 (1973); Helenius et al., Biochim. Biophys. Acta, 415, 29-79 (1975); and Morein et al.. Nature, 276, 715-718 (1978)], Skelly et al. prepared -HBsAg polypeptide vaccine in micellar form from the serum of infected chimpanzees using Triton X-100 ^® at 0.1% and at 2% [J. gen. Virol., 4_4, 679-689 (1979); and Nature, 290, 51-54 (1981)]. Methods for production of recombinant HBsAg polypeptides generally employ 0.1% or 2% concentrations of Triton X-100 ^® as in Skelly et al, J. gen. Virol. , 44, 679 (1979) [Rutter et al, European Patent Application No. 62,574; Miyanohara et al., European Patent Application No. 105,149; Valenzuela et al, in "Modern Approaches to Vaccines," Chanock et al. eds., 209-213, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York (1984); and Valenzuela et al.. Biotechnology, 3_' 317-320 (1985)], 0.1% (v/v) of a polyoxyethylene (9) octaphenol, nonionic detergent [Rutter et al.,European Patent Application No. 72,318], or are not reported as employing any detergent (Hitzeman et al., European Patent Application No. 73,657). However, there is no indication in the art that 0.1% and 2% are not preferable concentrations for the extraction of recombinant HBsAg from host microorganisms.

According to Adamowicz, et al., U.S. Patent No. 4,522,809, 0.02% to 0.2% of a nonionic detergent may be used to contact a chloroform-aqueous solution containing a viral suspension in order to obtain heavy subunits of influenza virus. Nevertheless, no suggestion is made that the use of a different concentration of nonionic detergent might result in a high yield of subcellular particles.

HBsAg may be isolated from serum as a mixture of particle types: small, spherical particles about 20 n in diameter; tubular particles about 20 nm in diameter; and large spherical particles about 42 nm in diameter with a core. A method for conversion of these particle types into a uniform spherical particle of 18 to 22 nm in diameter involves heating affinity-puri ied HBsAg for from 5 to 120 minutes at 40 to 80 degrees C. (preferably 60 degrees C.) in an isotonic. sodium chloride solution: having' a pH ranging from 5 to 9 and prferably being 7.2; and containing 0.05% to 5% (preferably 0.5%) of a surfactant. Surfactants useful according to this method include alkalai metal salts of bile acids, alkalai metal salts of lauroyl sarcosinic acid, polyoxyethylene alkyl phenols containing an average of 7-10 molecules of oxyethylene and polyoxyethylenesorbitan monoalkylester containing an average of 20 molecules of oxyethylene capable of delipidation in isotonic sodium chloride solution at about neutral pH. HBsAg particles having a molecular weight of about 2,200,000 daltons are collected in an aqueous solution. The surfactant is removed from the aqueous solution of antigen particles by dialysis or gel filtration. Funakoshi, et al., U.S. Patent No. 4,113,712. Although this method may achieve some interconversion of forms of HBsAg, the reported yields of 4.2% to 6.4% and the requirement for heating the HBsAg particles do not suggest application to extraction

of recombinant HBsAg where yields in the tens of percentage points are routine and where heating is not preferable because the presence of proteolytic enzymes necessitates extraction at low temperatures. Furthermore, there remains a concern about the use of ionic detergents with respect to the reported potential for ionic detergents to denature proteins during extraction. Helenius et al.. Methods in Enzymology, 56, 734-749 (1979).

In Hitzeman, et al.. Nucleic Acids Res., 11, 2745-2763 (1983), a 1% solution of a nonionic detergent is used to extract monomeric HBsAg proteins remaining in a cell pellet after extraction of micellar particles by means of a glass bead mill. No suggestion is made by Hitzeman, et al. to use 1% nonionic detergent solution to extract HBsAg particles, even though a comparison of glass bead extraction with 0.1% Triton X-100 ^® -ext ^' raction is prospectively described as "interesting".

Thus there exists a need for improved methods for the extraction and isolation of HBsAg accompanied by enhanced micelle formation.

Summary of the Invention A yeast cell-lysis buffer according to the present invention enhances the recovery of subcellular particles. The cell-lysis buffer includes an aqueous solution of a non-ionic detergent at a concentration within the range from about 0.5% to about 1.0% by volume.

A method according to the present invention provides for the highly efficient extraction of subcellular particles from cells. The method involves suspending the cells in a lysis buffer comprising an aqueous solution of a non-ionic detergent at a concentration within the range from about 0.5% to about

1.0% by volume. The method also involves maintaining the cells in the lysis buffer at a temperature within a range from about 3°C. to about 8°C.

Brief Description of the Drawing

The figure illustrates the effect of the concentration of Triton X-100 ^® in a cell lysis buffer upon the recovery of HBsAg micellar particles from the lysis of yeast host cells.

10

Detailed Description

In an examination of HBsAg recovery from Saccharomyces cerevisiae (S. cerevisiae) , a cell-lysis m m buffer was prepared as follows:

50m M Tris [tris(hydroxymethyl)aminomethane], • pH.8.0 lm M EDTA 150mM NaCl 20 10% (v/v) glycerol

0.1% (v/v) Triton X-100 ^®

Cells of S_^ cerevisiae expressing HBsAg according to Bitter, et al., Gene, 32, 263-274 (1984) 25 were suspended in approximately 2 volumes of lysis buffer to obtain a suspension having an ODg _Q0 (optical density at 600 nm) of 10.

Experimental runs are performed in a glass bead mill, of the sort available from Willy A. Bachofen

30 AG Maschinenfabrik, Basel, Switzerland, under the trademark Dyno ^® mill KDL. This mill operates by rapidly agitating a mixture of glass beads and cells suspended in a liquid medium. The extent of cell lysis is controlled by varying the agitation speed of the 5 beads. The extent of cell lysis may also be varied by varying the residence time of the cell suspension in the

mill. The operating chamber of the glass bead mill has a cooling jacket for controlling the temperature of the operating chamber containing the glass beads.

The amount of HBsAg particles released into solution were measured by a radioimmunoassay as available under the name Ausria™ Assay from Abbott Laboratories, North Chicago, Illinois. The amount of protein was also measured by the Biuret method, as an indication of cell breakage.

Recovery of HBsAg from freshly harvested yeast cells was compared to recovery from frozen yeast cells. The recovery from these two types of cells was very similar so that frozen cells were used thereafter for convenience.

It was observed that the amount of released HBsAg increased with decreasing cell concentration in the mill. This is contrary to published information about the mill which indicates that a higher specific release of enzymes and proteins is obtained with increasing cell concentrations. In addition, the highest reproducible recovery of HBsAg particles was at a level of only 30% of the value obtained from the control procedure.

Examination of the recovery of HBsAg per gram of cell versus the concentration of cells in the mill resulted in the determination that there was a factor limiting the amount of HBsAg particles produced.

In order to determine the identity of the limiting factor, a study of buffer components was undertaken. The following Examples disclose the results of that examination and provides a cell-lysis buffer useful for the production of subcellular particles. In Example I, the buffer and method are utilized in a test tube scale model. In Example II, the method and buffer

are employed in a batch process. Example III describes a preferred composition for the buffer and applies the buffer and method to a continuous process.

Example I

In a study of cell-lysis buffer components, the concentration of Triton X-100 ^® in the cell-lysis buffer described above was varied between 0% and 2% by volume and processed as described below.

10 Specifically, HBsAg-expressing S^ cerevisiae cells, according to Bitter, et al., supra, were suspended in 300 ml of a fermentation medium at O.D.10 (pH 4.45) and were centrifuged in a Beckman J6B centrifuge, available from Beckman Instruments, Mountain

15 View, California, at 4000 rpm (approximately 5000g) for 10 minutes at 4°C.

The resulting pellet was resuspended in 30 ml of 10% glycerol, 50mM Tris, ImM EDTA, 150 mM NaCl, pH

20 8.31. The cells were dispersed, and a 2.5 ml sample of cell suspension was added to each of 11 test tubes containing one of eleven batches, each batch containing a different concentration of Triton X-100 ^® as shown in Table I. Dilutions were made with

25

50mM Tris ImM EDTA 150mM NaCl

30 10% glycerol

at 4°C. Acid-washed glass beads were added up to the miniscus of the liquid. Each tube was covered with parafilm and vortexed at top speed in five runs of 15 m m seconds each. The sample was cooled between runs. The sample was mixed with two milliliters of 10% glycerol,

50mM Tris, ImM EDTA, 150mM NaCl pH 8.0 and centrifuged at 2000 rpm for 15 to 20 minutes. The supernatant was removed for assay.

A Biuret protein assay and a radioimmunoassay (Austria™ Assay, Abbott Laboratories, North Chicago, Illinois) were run for a sample of each batch with results as shown in Table I.

TABLE I

HBsAg

10%TX 100 Biuret ( ^μ g/g

In TEN & TEN & Protein HBsAg Wet

Batch Glγc.(yl) Glyc.(μl) OD540 (mg/ml) (ng/ml) Yeast)

4 500 0 .080 2.80 103 41.2 1

5 250 250 .082 2.87 220.5 88.2 •P-

6 125 375 .078 2.73 196 78.4 1

7 62.5 437.5 .082 2.87 183 73.2

8 31.25 . 468.75 .080 2.80 117 46.8

9 0 500 .083 2.91 <62 <24.8

10 500 0 ^' .069 2.42 72 28.8

11 250 250 .073 ^" 2.56 187 74.8

12 125 375 .080 2.80 198 79.2

13 62.5 437.5 .077 2.70 130 52.0

14 31.25 468.75 .075 2.63 112 44.8

As illustrated in the Figure, small variations in the concentration of Triton X-100 ^® made a large difference in the amount of antigen released. In fact, in the study of buffer components referred to above, the largest sample to sample variation between runs that were otherwise identical occurred with different concentrations of Triton X-100 ^® . The results shown in the figure indicate that the 0.1% concentration of Triton X-100 ^® commonly used is not adequate at high cell concentrations, and the optimal concentration of Triton X-100 ^® is between 0.5 to 1.0% by volume. Cell-lysis buffers having a Triton X-100® concentration within this range exhibited 2-fold greater release of antigen

EXAMPLE II

In an application of the method of Example I to a batch wise operation in a glass bead mill, frozen cell pellets from HB ^' sAg S_^ cerevisiae cell paste prepared according to Bitter, et al., supra, were prepared as indicated in Table II. The cell pellet was resuspended in 4 volumes of 50 M Tris buffered containing lOmM EDTA, at pH 8.0 and at 4°C by using a mixer. The resulting cell suspension was centrifuged in a Beckman J6B Centrifuge,as above, at 5000g (i.e. at 4000 rpm in a JS 4.2 rotor available from Beckman

Instruments, as above) for 5 minutes at 4°C. The weight of the pellets obtained is given in Table II.

Each pellet was resuspended in 1.5 volumes of 1% (v/v) Triton X-100 ^® in the cell-lysis buffer as described above. The cells were dispersed with a mixer and the pH of the solution was adjusted to 8.0± 0.1 with 50% NaOH as needed.

After passage through the glass bead mill, as described above, samples from three runs were assayed as indicated in Table II. A small scale lysis as described in Example 1, runs 8 and 14, was run as a control.

TABLE II

HBsAg in

HBsAg i HBsAg in Soluble

OD 10 Soluble Fraction

Test Fraction of

Tube of Dynomill

Triton Cont Dynomill Lysate

Run X-100 ^® Cells Lysis Lysate (μg/g of % of

No. (v/v%) [Wet Kg)/L] (μg/OD- -L) ^• (yg/OD-L) wet cells) Control

1.0 0.33 185 263 70 142

0..9 0.32 129 276 90 214

0.9 0.30 121 323 103 267

Thus, recovery of 150% of the control values obtained in test tube runs may be obtained by the batchwise method of lysis.

EXAMPLE III HBsAg cell suspensions were prepared as in Example II with two of five samples of cells being suspended in 1% (v/v) Triton X-100 ^® and three of five samples being suspended 0.5% (v/v) Triton X-100 ^® .

A glass bead mill was primed by pumping 2 liters of cell-lysis buffer, described above,through the reaction chamber while chilling the mill 5° ±3°C. The mill was turned on and cold cell suspension was pumped into the mill. One retention volume of lysate was discarded while the remaining lysate was collected for further processing.

The retained lysate was centrifuged in a Beckman J6B Centrifuge at 5000g (4000 rpm) in a JS 4.2 rotor for 30 minutes. Assays were performed on the supernatant and the results are given in Table III. A small scale lysis as described in Example 1, runs 8 and 14, was run as a control.

TABLE III

HBsAg in

HBsAg in HBsAg in Soluble

OD 10 Soluble Fraction

Test Fraction of

Tube of Dynomill

Triton [Cells] Lysis Dynomill Lysate

Run X-100 ^® (Wet Control Lysate (mg/Kg of % of No. (v/v%) Kg/ml) μg/OD-L) (μg/OD-L) wet cells) Control H- ^a

00

1 1.0 0.21 475 499 70 105 2 1.0 0.21 525 985 390 188 3 0.5 0.21 512 918 443 179 4 0.5 0.20 175 472 181 270 5 0.5 0.30 236 558 214 236

All of the runs in Table III were made at a 15-minute residence time in the glass bead mill. The residence time for maximum release of antigen was determined to be between 10 and 15 minutes. In the continuous runs reported in Table III, yields equal to 100% of the control lysis results were achieved.

Because subsequent purification steps are facilitated by having as low a concentration of Triton X-100 ^® as possible, a concentration of 0.5% (v/v) is currently preferred. The concentration of EDTA is preferably increased from ImM to lOmM in order to prevent proteolysis in the lysate. Neither of these modifications significantly affects the yield of HBsAg. Accordingly, the currently preferred buffer for cell-lysis is:

50mM Tris, (4°C) pH 8.0 lOmM EDTA 150mM NaCl 10% glycerol

0.5% (v/v) Triton X-100 ^®

Although the present invention has been described in terms of a preferred embodiment, it is understood that modifications and improvements will occur to those skilled in the art. For example, although Triton X-100 ^® has been used as a nonionic detergent in the methods and buffers according to the present invention, it is expected that other nonionic detergents such as Tween 80, Tween 20 and NP-40 will prove useful as well. Similarly, although the subcellular particle isolated in the examples above was a HBsAg particle, it is expected that other subcellular particles may be extracted from cells using the buffer and method according to the present invention with slight modifications.

Accordingly, the present invention is intended to encompass any method or buffer which comes within the scope of the invention as claimed.