HEPATITIS A VIRUS CULTURE PROCESS

BACKGROUND OF THE INVENTION

The invention pertains to a device and process for improved cultivation of surface-adherent animal cells; a method for producing viral vaccines, natural or recombinant proteins.

A general challenge in adherent animal cell culture is the supply of large amounts of surface area for cell attachment (a prerequisite to cell growth and proper function for many cell lines), preferably within a single vessel, and preferably with as high a surface to volume ratio as possible. This has been provided in the past by many different packed bed reactor designs, including hollow fiber reactors, packed beds of spheres, packed beds of randomly oriented fibers, and porous ceramic monoliths. The goal in all of these is to provide as much surface area as possible in a single, simple, scaleable reactor. However, these reactors have met with only limited success; no Un¬ licensed human vaccines or therapeutics are known to be produced in these systems.

All of the above reactors have documented difficulties with maintaining nutrient supply to the cells in the reactor. This arises due to two factors. First is the configuration of the reactor itself. In all of the reactors, with the exception of the ceramic monoliths, the packing is randomly configured. Hollow fiber reactors rely on diffusion and Starling flow to supply medium flow across the cell compartment [J. M. Piret (1989), B.c.D Thesis, Dept. of Chem. Eng., Mass. Inst. Technology, August 1989] . Diffusive nutrient penetration is only adequate if the depth of the cell accumulation on the fibers is controllable and uniform, which it is not in these reactors. Starling flow decreases as cell growth occurs, due to increased flow resistance offered by the cell growth. This reduces mixing within the cell growth chamber. Randomly packed beds of beads or fibers rely on forced convection, and are both prone to channeling through paths of least hydraulic resistance, bypassing the areas where surface area is most

dense, and hence likely to contain cells. Porous ceramic monoliths are better in this respect, but these suffer from the second factor, biofouling. As the cells grow on the attachment surface, they can constrict and occlude the paths where medium flow occurs [J. E. Putnam et al., Ann. Mtg. Soc. Ind. Microbiol., Orlando, Florida, Aug. 1, 1990]. In the case of the ceramic monolith, cell growth results in restriction of the channel, such that increased hydraulic resistance to flow is offered. The medium then preferentially flows to other parallel channels, and the result is that the channel with the heaviest cell growth receives the least medium. In extreme cases, cell growth can result in catastrophic reactor plugging. In the case of randomly packed beds of spheres or fibers, the irregular packing allows the cell growth to constrict flow in the areas where cell growth is most vigorous, thus depriving the cells of nutrients and shifting flow away from growing regions of the reactor. The effect of cell growth on the hollow fiber reactor has been mentioned above.

The use of solid static mixer elements in cell and virus culture is known, see Grabner and Paul, U.S. Patent 4,296,204.

The use of a mesh for culturing primary tissues comprised of several cell types is found in U.S. Patent 4,963,489 and 5,160,490. No configuration of the mesh is suggested, indeed in one example (Example 13, Pat. 5,160,490) the meshes are directly stacked upon each other in a stagnant culture, an unsatisfactory situation for optimal cell nutrition. A testing apparatus made of mesh is claimed in 5,160,490. The tissue resulting from a mesh culture of stromal fibroblasts is claimed in 4,963,489.

Growth of limulus amebocytes on a stainless steel mesh is suggested by Friberg et al. [In Vitro, 28A, 215, (1992)], where it is rated as a poor substrate for cell growth. Stainless steel mesh is also used in Epstein et al.. [In Vitro, 25(2), 213, (1989)], for the production of CHO cells secreting a recombinant protein. Here, the purpose of the mesh is to expand the surface area available in a single culture vessel. The mesh is not configured in such a way as to positively influence the nutrient supply to the cells.

The use of a fibrous bed reactor has been suggested by Perry and Wang, [Biotechnology and Bioengineering, 34, 1-9, (1989)]. Those authors suggest use of randomly-packed fiber beds. A subsequent article by Chiou, Murakami, Wang and Wu, [Biotechnology and Bioengineering, 37, 755 (1991)] employs woven fiber mats laid upon each other for the bed. No provision was made for preventing channeling of the fluid flow, but state that it does not occur on a macroscopic basis, as estimated by radial measurements of cell concentration through their reactor, and by virtue of cellular productivity estimates compared to monolayer systems. The fiber mats are in close contact to one another, however, potentially allowing cell overgrowth and maldistribution of nutrients to occur. In any case, no provision for control of nutrient flow to the cells was made.

In vitro growth of hepatitis A virus is known. In 1973, Feinstone et al., [Science 182. p 1026] identified the etiologic agent of infectious hepatitis, later known as hepatitis A virus, using immune electron microscopy. In vitro culture of hepatitis A virus (HAV) was first reported by Provost et al [P.S.E.B.M. 160, p213, 1979] according to a process whereby liver from HAV infected marmosets was used as an inoculum for liver explant culture and fetal rhesus kidney (FRhK6) cell culture [U.S. Patent 4,164,566]. In a later invention, direct inoculation of a HAV, which had not been previously passaged through a subhuman primate, was successfully used to initiate in vitro propagation of HAV [Provost et_al., P.S.E.B.M. 167, p201 (1981); U.S. Patent 5,021,348].

From this work, attenuation of HAV through in vitro culture was demonstrated. In addition, it was demonstrated that upon repeated passage in vitro. HAV cultures became more productive and replication rate accelerated as the virus became adapted to the cultured cells. A further development was the demonstration of protective efficacy of both the live attenuated virus [Provost; et. al.. J. Med Viol. 20, p 165 (1986)] and the formalin inactivated HAV [U.S. Patent 4,164,566; U.S. Patent 5,021,348; Provost et. al.. in Viral Hepatitis and Liver Disease. p83-86, 1988 - Alan R. Liss, Inc.]. From the foregoing

work, it has become clear that either an inactivated or attenuated, immunogenic HAV are possible vaccine candidates. However, a reproducible, commercial scale process for production of high purity antigen is needed if a safe HAV vaccine is to be commercially available for use in humans.

Various methods have been described to culture HAV for vaccine production. Thus, Provost et al. (US 5,021,348) described a process whereby, in a preferred method, a cell culture of MRC-5 cells was infected with HAV. According to that disclosure, the virus and cells are grown according to conventional methods in monolayer. In US Patent 4,783,407, HAV was grown in Vero cells (a type of primate kidney cell). In US Patent 4,301,209, high titer HAV production in a hollow fiber capillary unit was described. In US Patent 4,412,002 a process whereby HAV was isolated from persistently infected cells was described. In EP 0 302 692, HAV culture in roller bottles was described. In all of these systems, the large scale production of HAV required for a commercial process was not feasible or was severely limited by the amount of surface area available for cell sheets to be established for HAV infection. Cell culture devices are now available which offer substantial increases in the availability of surface area for attachment dependents cell growth. Thus, NUNC cell factories, which are multilamellar plastic units, and COSTAR cubes are useful, commercialy available devices for moderate to large scale production of cell sheets and HAV production. In addition, in US Patents 4,296,204 and 4,415,670, motionless mixers as cell propagators were described. Thus, the options for large scale HAV production have been expanded. However, even within the foregoing apparatuses, the volume to surface area ratio is unacceptably large for a commercial scale operation.

As noted above in US Patents 4,963,489 and 5,160,490, Naughton and Naughton described a three-dimensional cell and tissue culture system in which cells derived from a desired tissue-source were inoculated and grown on a pre-established stromal support matrix. The stromal support consisted of fibroblasts grown to subconfluence on a

three-dimensional matrix. The three-dimensional matrix was comprised of nylon or other polymer, cat gut sutures, cellulose, gelatin, or other biological polymer. The purpose of that invention was to produce a biomimetic system for ex vivo production of tissues on which the toxicity of test compounds could be ascertained. Use of a woven fiber sheet was to provide an ordered three-dimensional arrangement for tissue formation by multiple cell types which were co-cultured; no special attention was paid to nutrient supply or reactor configuration.

The present invention arranges surface area for cell growth into a high area per volume configuration which ensures that during cell growth, nutrition to the cell sheet does not deteriorate due to local biofouling. In the current invention, cell growth actually serves to recapture the plug flow behavior of solid surface static mixer reactors, thus improving nutrient supply, while providing a larger surface area for growth than possible for solid surface reactor, by a factor of two or greater. It thus provides advantages over solid surface reactors by virtue of arearvolume ratio, while possessing nutrient supply advantages over previous reactor designs intended to provide high area:volume. In the present invention, a mesh is used to provide a high surface area to volume ratio in a bioreactor. Attachment dependent cells are grown on the surface area thus provided and are infected with HAV or other viruses to produce a very large-scale culture process for HAV or other viral vaccine production.

SUMMARY OF THE INVENTION

A mesh cloth, or porous solid of known thickness is fabricated into the shape of a static, or motionless, mixer element (cylindrical or otherwise), wherein the cloth or porous solid serves as a growth surface for animal cell cultures. This arrangement allows for: 1 ) superior surface area for cell growth per unit volume of reactor; 2) superior cell growth on this surface due to the regular fiber or pore spacing; 3) known and controlled cell nutritional environment due to the uniform bed packing arrangement into a static mixer configuration; 4) avoidance of "dead zones" in the reactor due to cell growth; 5) re-use of

the elements due to materials of construction and cleanable configuration. Mixing in the reactor improves with cell growth as opposed to other cell culture reactors. The invention is a new process for growing cells, a new apparatus in which to grow the cells, and a new method of use for existing mesh static mixer devices.

A process is provided whereby culture of HAV in a large scale stationary bioreactor with a high surface area to volume ratio is provided by growing HAV infection susceptible cells on a mesh. The HAV thus produced may be recovered by detergent harvest of the virus and down-stream processing to produce an HAV vaccine. In a preferred embodiment of the invention, the bioreactor comprises mutiple stainless steel wire mesh elements disposed in a circulatory loop in which medium perfusion can occur. This embodiment results in production of cells to about lθ6 cells/cm^ for vaccine production.

BRIEF DESCRIPTION OF THE FIGURES

Figure 1. MRC-5 cell growth on glass, polystyrene, and 316 stainless steel. Cell surface concentration is plotted as a function of time. Cells were inoculated on coupons of each material at a density of approximately 10,000/cm^ and a coupon was sacrificed daily for cell counts. Cells on glass coverslips and 316 stainless steel coupons were cultivated in 150 cm^ petri dishes containing 90 mL of medium (0.6 mL/cm^); cells grown in 25 cm^ T-flasks were cultivated with 7.5 mL of medium (0.3 mL/cm^).

Figure 2. MRC-5 cell growth and residual glucose concentration for cells grown on 316 stainless steel coupons. Cell surface concentration (cells/cm^) and glucose concentration (mg/cm^) are plotted as a function of time (d) for cells grown on 25 cm^ stainless coupons.

Figure 3. Light micrograph of MRC-5 cells growing on stainless steel gauze. Cells were inoculated at 5,000 - 10,000 cells/cm^ and cultivated in excess of 2 weeks. Magnification 40X.

Figure 4. Viable and non-viable staining of MRC-5 cells grown on 316 stainless steel gauze. Cells were stained with fluorescein diacetate (FDA) and ethidium bromide (EB) according to the protocol described in the Material and Methods section of below, and then viewed using a scanning laser confocal microscope; the material was scanned in 57 μm increments. Figure 4A corresponds to fluorescence from fluorescein, and indicates viable cells. Figure 4B is fluorescence due to ethidium bromide staining of DNA, demonstrating the near complete viability of the cells. The positive control for ethidium bromide is shown in Figure 5.

Figure 5. Ethidium bromide stain of non-viable MRC-5 cells. Cells grown on 316 stainless steel gauze were made non-viable by treatment with 100% methanol. After rinsing well with PBS and allowing to dry cells were stained with FDA/EB to serve as a negative control for FDA and a positive control for ethidium bromide (pictured). No fluorescence due to fluorescein production was observed for the negative control.

Figure 6. Planting efficiency is plotted for each element in a gauze static mixer reactor for protocols employing gravity sedimentation, or medium recirculation. Elements are numbered 1 through 5, with 1 being the top element and 4 or 5 being the bottom element (4 elements were used in gravity sedimentation experiments and 5 for recirculation experiments). The planting efficiencies presented here are the percent of the total planted; overall planting efficiencies were on the order of 90%. Data for recirculation is an average of 3 experiments while that for gravity sedimentation is from a single experiment. Recirculation velocities were approximately 6 cm/h.

Figure 7. Glucose uptake rates (GUR) as a function of time for reactors containing solid titanium and gauze 316 stainless steel elements. GUR on a per cm^ basis (μg/cm^-d) against time (d) is shown in Figure 7A

while GUR on a per reactor volume basis is plotted in Figure 7B. The data in Figure 7 A is dependent on the estimated surface areas of the elements while Figure 7B embodies no estimation since both reactors are the same volume.

Figure 8. The percent of total protein observed in successive lysates for the solid and gauze static mixers. The protocols for lysates 1-4 are described in the text. By microscopic observation, solid elements were free of cell debris while the gauze elements contained considerable cell debris after the fourth lysis.

Figure 9. Glucose uptake rates as a function of time for a 316 stainless steel gauze static mixer reactor. The reactor was infected with an MOI of 1 on day 7 and harvested on day 28.

Figure 10. The percent of total protein and HAVAg in successive lysates for the gauze static mixer shown in Figure 9. The protocols for lysates 1 - 4 are described in the text. The total protein of lysate 1 was artificially high due to inefficient PBS washes.

Figure 1 1. The effect of MOI on cell growth and glucose consumption. Cells were infected with an MOI of 0, 1, and 10 with HAV CR326F P28. Differences in cell density (HA) and residual glucose (1 IB) are not observed until 8 days post-infection for this system. A complete medium refeed was performed on day 8.

Figure 12. Cellular protein measured against a BSA standard versus number of cells. The best fit line is described by the equation: Y(IO^) = -0.09 + 0.42X, r = 0.99. This Figure demonstrates that a single curve provides a means to measure MRC-5 cell number by total protein, regardless of the "state" of the cells.

Figure 13. Figure 12 is replotted, indicating the time of each data point.

Figure 14. Conversion of total protein to a cell number closely resembles the cell density obtained using a Coulter Counter. It is believed that the discrepancies are due to the small volumes of lysis buffer used in harvesting the T-25s (2x0.08 mL/cm^ was used for consistency).

DETAILED DESCRIPTION OF THE INVENTION

The reactor disclosed in this invention is useful for propagation of cells for the production of cell culture derived biologicals. Molecular applications include any chemical (e.g. - tPA, erythropoetin) or chemical assemblies (e.g. - Factor VIII V on Willebrand's factor complex) which can be produced by, or modified by, an adherent animal cell culture. The cells are grown on the reactor surface to an optimal and high surface density, and induced to produce the product by means well know in the art. The product is collected by harvesting the supernatant medium, by recovering product from the cells by an in situ treatment, or by removing the cells from the reactor and recovering product.

Vaccine applications include any virus which can be cultured on an anchorage-dependent animal cell culture, and can be recovered from the reactor (e.g. - Hepatitis A, Mumps, Measles, Rubella, poliovirus, rotavirus). Here, cells would be grown to an optimal and high surface density, and would be infected with virus. The product virus would be collected as above from either supernatant medium or cells.

Cellular applications include growth on the reactor and subsequent recovery for use (e.g. - blood cells from bone marrow stem cell cultures). Here, cells may be grown to an optimally high surface density, and induced to produce product cells which would then be collected from the reactor or from supernatant medium.

In one embodiment of this invention, hepatitis A virus (HAV) variant P28 of strain CR326F was used to infect MRC-5 cells grown in the apparatus of this invention, for merely illustrative

purposes, and the production material was cultured at P29. P28CR326F is an attenuated HAV strain. Other strains and/or serotypes of HAV are encompassed by this invention, including HAV strains that can be attenuated by conventional techniques known in the art. Other suitable cell lines for HAV propagation include Vero, FL, WI-38 and FRhK6 cells. These and other systems for HAV propagation in cell cultures are discussed in Gerety, R.J. "Active Immunization Against Hepatitis A," in Gerety, R.J. (ed.) Hepatitis A Academic Press 1984, pp. 263-276; and Ticehurst, J.R., Seminars in Liver Disease 6, 46-55 (1986). In principle, any cell line such as any human diploid fibroblast cell line, can serve as a host cell for HAV provided that it is susceptible to HAV infection. The preferred cell line is MRC-5.

The apparatus of this invention utilizes a gauze or mesh containing static surface mixer. In a preferred embodiment of this invention, MRC-5 cells are infected at a multiplicity of infection (MOI) of HAV sufficient to achieve efficient cell culture infection. An MOI of about 0.1-1 is acceptable. Stock seed is conveniently generated by using HAV from the supernatant fraction of an SSR incubated for about 28 days.

In a preferred embodiment, HAV is cultured on MRC-5 cells grown on stainless steel gauze. According to this embodiment a bioreactor is used wherein mesh elements, made from any suitable, nontoxic material on which cells can grow, such as stainless steel, titanium, plastic or similar material which can be provided in a regular three-dimensional array, preferably 316 stainless steel mesh cylindrical elements such as is available from Koch Engineering Company, Inc., MBR Bio Reactor AG, or Sulzer Biotech Systems, wherein each element of meshwork is oriented at an angle up to 90° to the preceeding element to increase mixing. Cells are then seeded and allowed to attach to the vast surface area provided by the mesh. Once the cells are well established in culture, they can be fed at a steady perfusion rate to maintain gas exchange and elimination of waste products. At an appropriate cell density, the cells are infected with HAV as further described below.

The advantage of the mesh elements is that the fibrous bed allows more surface area per unit volume. Geometrically precise mesh allows uniform growth of fibroblasts across the surface, to known and constant depths, allowing control of nutrient delivery. The static mixer configuration controls the proximity of the growth surfaces to each other such that natural biological growth and "fouling" of the reactor is avoided. This is a serious problem encountered in random-packed bed reactors, where the non-uniformaties of the packing allows localized regions of cell growth which then cause a loss of nutrient flow to that region and subsequent cell starvation. With the static mixer configuration, cell growth to fill the interstices in the mesh diverts the nutrient medium flow so that it recaptures the flow pattern of a solid static mixer, a configuration well known for its uniform flow properties.

It will be understood that the scope of the present invention encompasses, in addition to the passage 18, P18 or p28, of strain CR326F' of HAV, any other HAV variant or strain, whether attenuated or virulent as well as other viruses which can be cultured on anchorage dependent cells. Attenuated variants or strains may be isolated by serial passage in cells, animals, or by other methods. See, for example, Provost, PJ. et al. Proc. Soc. Exp. Biol. Med. 170,8 (1982); Provost, P.J. et aL J. Med. Virol. 20, 165 (1986): U.S. Patent 4,164,566 and 5,021,348 for details on attenuation. The culture method of the present invention is readily and easily adaptable to attenuated or virulent HAV strains.

The HAV is allowed to replicate in the SSR to a peak of virus production. This takes about 21-28 days, during which time nutrient medium is constantly circulated through an extemal loop which provides oxygenation. The cell culture medium may be any medium which supports active growth of MRC-5 cells and HAV replication. Preferably, the culture medium is replenished and removed at a constant rate between about 0.010 and 0.3 mL/cm^/day.

Thus, in a preferred embodiment of this invention, MRC-5 cells are grown in cell sheets in a SSR comprising stainless steel gauze

or mesh elements, infected at a multiplicity of infection (MOI) of HAV of about 0.1-1.0, and the HAV is allowed to replicate to a peak of virus production. This takes about 28 days, during which time nutrient medium is replenished and circulated through the culture vessel and an external or internal loop which provides oxygenation. The culture medium may be any medium which supports active growth of MRC-5 cells and HAV replication. We have found that an iron supplemented calf serum medium is preferred for this purpose.

(b Harvesting the cultured HAV.

Following culture for about 28 days with nutrient replacement, the culture medium is drawn off and the HAV is harvested. In the past, this has been accomplished by scraping followed by freeze thawing and optionally sonicating to optimize release of the cell-bound virus. In the process of this invention, cell bound HAV is liberated into a minimal volume of harvest solution. Preferably, the harvest solution contains a component effective to render the cells permeable to HAV. Such components are known in the art. Preferably, a detergent such as Triton X-100, NP-40, or an equivalent is supplied at the lowest effective concentration possible, to facilitate later removal. Provision of about 0.1 % Triton X-100 has been found adequate to achieve efficient extraction of HAV from cells. The advantage of using detergent extraction is that the geometry of the SSR culture vessel and the meshwork does not permit harvest by mechanical means, except to the extent that HAV is found in the culture supernatant, which can be drawn off and the HAV concentrated and recovered. However, the proportion of HAV found in culture supernatant is generally only a small fraction of the total HAV recoverable from the cells.

(c) Harvested HAV Treatment.

Once the HAV has been harvested in a substantially cell- free form, from the culture supernatant, from the cells, or both, it is desirable to concentrate the HAV to facilitate down-stream processing. The HAV produced in culture according to the instant invention may be

purified according to methods known in the art or it may be directly used as a vaccine if attenuated. It may also be inactivated according to methods, known in the art, primary among which is formalin inactivation. For details on these steps known in the art, see for example US Patent 4,164,566; 5,021,348; EP 0 302 692; and USSN 07/926,873, filed on 8/10/92. .

(d Vaccine Inactivation and Formulation.

Additional processing steps of conventional and well known character are or may be needed to prepare purified HAV capsids for vaccine use. For example, treatment with formalin, sterile filtration and adsorption to carriers or adjuvants are the typical basic steps for preparing a formalin-inactivated vaccine. See, for example, Provost, PJ. et ai Proc. Soc. Exp. Biol. Med. 160, 213 (1979); Provost, P.J. et al. J. Med Virol. 19,23 (1986). HAV can be inactivated by heat, pH changes, irradiation, treatment with organic solvents such as formalin or paraformaldehyde. Typically, HAV inactivation is carried out at a 1/4000 ratio of formalin. The inactivated HAV is then adsorbed or coprecipitated with aluminum hydroxide to provide adjuvant and carrier effects.

Because the dosage required for immunological efficacy of this product is very low, it is convenient to complex or adsorb the inactivated HAV onto a carrier. Aluminum hydroxide has been found to be quite acceptable for this purpose as it forms a tight complex with the HAV and prevents loss of the virus onto the walls of the container. Efficacy of an inactivated HAV vaccine has been shown [New England J. of Med. 327: 453-457, (1992)].

The studies disclosed herein demonstrate that static mixer elements constructed of stainless steel gauze are able to support intersitial and multilayer growth of MRC-5 cells. Multilayers are demonstrated to be viable, with no indication of necrosis. Infection studies indicate that these multilayers are susceptible to hepatitis A infection and in a non-optimized system produce antigen in excess of what is currently realized using solid supports. With microcarrier

culture, cell clumping may result in ill-defined multilayers; the gauze static mixer provides well defined multilayers due to the crafted geometry and hence uniform nutrient delivery, and therefore nutrient limitation and waste accumulation pose less of a problem. Cell growth within the interstices of the mesh closes off the interstices, and creates a plug flow pattern similar to that of solid sheet elements. Therefore, the system reaps the mixing benefits of static mixer technology and leads to predictable and uniform nutrition to the cells. This is in contrast to other packed bed reactor types such as randomly packed beds of fibers, packed beds of glass beads, hollow fiber reactors, or ceramic monolith reactors. In these types of reactors, cell growth causes deterioration of the medium flow patterns and creates poorly nourished pockets of cells. Furthermore, these elements are chemically cleanable for reuse in cell culture; this is desirable for automation in manufacturing environments. In summary, a reusable reactor containing stainless steel gauze static mixer elements embodies great potential for the cultivation of MRC-5 cells and the production of hepatitis A virus, other viruses, recombinant proteins and animal cells expressing such proteins.

The following examples are provided to exemplify specific embodiments of this invention, but the examples should not be construed as the only mode of executing this invention.

EXAMPLE 1 HEPATITIS A VIRUS STOCK SEED MANUFACTURE

A large-scale procedure for virus seed production involves the infection of MRC-5 cell monolayers in 6000 cm2 NUNC CELL FACTORIES. MRC-5 cells are grown in the factories to confluent monolayers, and then infected with virus at an MOI of 0.1. Following infection the cells are incubated for 28 days with weekly replacement of medium containing 10% v/v fetal calf serum. It has been found that high concentrations of serum, 2 to 10% v/v allow greater virus production than low levels, 0.5 to 2% v/v. At the end of this cycle the supernatant fluid contains large amounts of virus, in this example 107.3

TCID50 per milliliter, which is harvested directly from the NCF, without cell lysis, and used as the source of virus stock seed. In this manner, the large quantities of infectious virus necessary for manufacture are obtained by a method more reproducible and facile than roller bottles or flasks, or with mechanical harvest of the cells.

EXAMPLE 2

Harvest of Hepatitis A Virus from NUNC CELL FACTORIES or Static

Surface Reactors bv Triton Lvsis

Neither the NUNC CELL FACTORY nor the Static Surface Reactor is amenable to mechanical scraping for release of MRC-5 infected cell sheet. The cell-associated Hepatitis A virus was recovered using Triton lysis and purified using a novel process. The solution obtained with the Triton lysis procedure was much more dilute than the preparation obtained from mechanical scraping of roller bottles. The Triton harvest was concentrated by diafiltration, followed by treatment with the absorbent, XAD, to remove residual Triton. Clearance of Triton to less than 10 mg/mL has been demonstrated using a Capillary Zone Electrophoresis in-process assay. The HAV was then precipitated with polyethylene glycol (PEG), extracted with chloroform:isoamyl alcohol. Contaminating DNA was removed using a DNA filter column, followed by anion exchange and size exclusion chromatography. The process involving both NCF propagation and the Triton lysis did not adversely affect the immunogenicity of the purified formalin-inactivated Hepatitis A vaccine.

EXAMPLE 3 HAV Production On Stainless Steel Gauze Static Mixer Elements

MATERIALS AND METHODS

For all subsequent examples, the materials and methods used were as defined in this example, unless otherwise specified.

Cell culture.

MRC-5 cells, at PDL < 40, were used in all studies. Cells were cultured in Williams E Medium containing 10% iron enriched calf serum, 2 mM glutamine, and 50 mg/L neomycin sulfate at 37° C. 0.1 % Pluronic F-68 was added to medium in studies employing a sparged stirred tank reactor. A cell density of 10,000 cells/cm^ was used to inoculate new culture surfaces. Cell counts were performed using a Coulter Counter and/or hemacytometer with Trypan blue staining to test viability.

Cell culture on metal coupons.

A solid or gauze metal coupon, ca. 5 cm x 5 cm, was placed in a glass, 150 cm^ petri dish and sterilized. After allowing to cool, 90 mL of medium was placed in the dish, and the cell inoculum was then added. The medium amount allowed for complete coverage of the metal surface. After 1 day at 37° C the metal piece was removed using sterile forceps, and placed in a new sterile petri dish (to discount cell growth on the glass surface) containing 90 mL of medium. The solid sheets were trypsinized by washing twice with phosphate-buffered saline (PBS), and adding 2 mL of trypsin to the metal. After observing the release of cells, the trypsin was pipetted gently over the surface several times to ensure complete removal.

Virus stock and infection.

Hepatitis A virus stock, CR326F P28, with a titer of 10 ⁸ TCID5o/mL, was used in all studies. MOI values reported here are calculated as TCID5o/cell and not PFU/cell. Cultures were infected by adding the virus stock to fresh medium and adding this to the culture. After infection cultures were incubate at 32° C.

Viability Staining on Gauze Coupons.

Cell viability was monitored by the use of fluorescein diacetate (FDA) and ethidium bromide (EB) staining. In viable cells FDA is

cleaved by esterases in the cytoplasm to yield free fluorescein that fluoresces green. EB rapidly intercalates into the nuclear DNA of non- viable cells and fluoresces red. Cells were labelled with both dyes simultaneously. The staining solution consisted of 100 μL of FDA stock (5 mg/mL in acetone) and 200 μL of EB stock (10 mg/mL in PBS) in 50 mL of PBS. The gauze material was then incubated in 5 mL of this solution for 5-6 min. After labelling, the support was washed twice in PBS (20 mL per wash) and taped between a slide and coverslip. The samples were then viewed using a BioRad MRC-600 confocal microscope.

Static mixer reactor.

Gauze (known as "E-pack") and solid sheet 316 stainless steel static mixer elements (SMV configuration) were obtained from Koch Engineering (Wichita, KS). All elements were approximately 2"x2", constructed of type 316 stainless steel or titanium. The surface area of the solid elements was estimated to be 1300 cm^ and the gauze elements were estimated to be 2300 cm^. The surface area to liquid volume ratio of these latter elements was approximately 26 cm^/mL (ca. 24 cmr/mL based on total reactor volume). Five elements were normally placed in a threaded 5 cm x 26.5 cm glass columns with PTFE end fittings. Elements were placed in a configuration with the mixing directions perpendicular in successive elements. The column with elements comprises the reactor for cell growth. Stirred tank satellite vessels coupled to electronic controllers were used to control the medium pH at 7.2 - 7.3 and the DO at 80-90%. Medium was recirculated from the stirred tank to the reactor by use of a peristaltic pump.

Reactor planting was accomplished by providing a low circulation rate to allow cell attachment. Inoculum was placed into the system and typically recirculated at 100 mL/min for 5 min to mix the cells and medium. Very little cell attachment occurred during this rapid recirculation period. The flow rate was then reduced to ca. 2 mL/min, which provided a superficial velocity through the reactor approximately equal to the sedimentation velocity of the cells. Velocities around the

cell sedimentation rate were found to be optimal for cell attachment rates and even distribution of cells through the reactor.

Cell lvsis procedure. Cells were lysed in a Triton lysis buffer which contained 0.1 %

Triton X-100, 10 mM TRIS, and 1 mM MgCl ₂ . T-flasks were washed twice with 0.3 mL/cm^ PBS, and lysed twice with 0.1 mL/cm^ lysis buffer for 1 hr. Protocols for lysing reactors is described in the text.

Analytical Methods

Cellular metabolites were measured using a Kodak Ektachem DT60 analyzer. Hepatitis A Viral Antigen (HAVAg) was assayed using a monoclonal antibody based enzyme immunoassay. Total protein of cell lysates was quantified using a Coomassie Blue Bio-Rad protein assay (Cat. #500-0001) against a bovine serum albumin standard.

EXAMPLE 4

Growth of MRC-5 cells on Solid Metal Coupons.

Static mixer work prior to this study was performed on titanium elements. It was desirable to investigate cell growth on surfaces of different alloys that, in addition to being less expensive than titanium, provide greater ductility for ease of static mixer manufacturing. Two grades of each stainless steel and Hastelloy were investigated: 316 and 316L stainless steel, and Hastelloy grades B and C. Metal coupons were inoculated with approximately 10,000 cells/cm2. After 4 days, the cells were trypsinized and the final cell densities recorded:

Material Cells/cm^

Titanium 1.4xl0 ⁵ 316 Stainless 1.6xl0 ⁵

316L Stainless 1.2xl0 ⁵

Hastelloy B 2.2xl0 ³

Hastelloy C 1.3xl0 ⁴

Growth on both grades of stainless steel was similar to that on titanium; cells did not grow on hastelloy B and grew poorly on hastelloy C. Differences in cell densities noted for the stainless steels and titanium should not be construed as evidence that one is superior; differences are likely due to assay variability.

Most of the data for Hepatitis A virus processes have been obtained for cells grown on plastic supports, e.g., polystyrene tissue culture flasks. For comparison, growth curves of MRC-5 cells on polystyrene, glass, and 316 stainless are shown in Figure 3. The cells grow comparably on all of the surfaces. The maximum specific growth rates are similar (0.034 h ^" * for plastic, 0.032 h ^~ * for glass, and 0.039 h ^~ * for 316 stainless steel). The lower final cell density obtained for plastic is likely due to nutrient limitation. Growth on plastic was performed in T-flasks while growth on glass and stainless steel was performed in petri dishes. Due to the difference in the volume of medium used per cm2 in each sytem, an imbalance in available nutrients results. Cell growth and the consumption of glucose for stainless steel coupons are shown in Figure 2.

EXAMPLE 5

Growth of MRC-5 on Gauze Metal Coupons

In an effort to maximize surface area per unit volume, metal gauze material was investigated in petri dishes for MRC-5 cell growth. For certain combinations of wire diameter and spacing, a piece of gauze with the same projected area as a solid coupon can present a greater overall surface area for cell growth. Sheets of gauze can then be formed or arranged into a reactor in such a way as to maximize the available surface area per unit of reactor volume. To test growth of cells on a gauze surface, MRC-5 cultures were conducted by the techniques described in the Materials and Methods section (Example 3).

Growth of MRC-5 cells was demonstrated on gauze constructed of either 316 stainless steel or titanium. Gauze coupons with wire diameters ranging from 100 μm to 400 μm, and weaves of 14 x 100, 120 x 110, 40 x 200, and 107 x 59 wires per inch were investigated.

Cells grew on all samples, to greater or lesser extents. In the early stages, cells grow on the wire surface. Shortly thereafter cells begin to bridge the interstices, especially at the corners (see Figure 3). With time the cells completely fill the interstices and form multilayers. By visual inspection, the gauze supports very high cell densities that are more than scaled by surface area. Direct cell surface concentrations are not available since multilayer growth on the gauze is not amenable to trypsinization. Preliminary data indicated that interstice size is an important consideration, as cells are unable to bridge large interstices, and gauze with small interstices reduces in the limit to the same surface as a smooth plate. Material of construction did not affect cell growth, and it is concluded that any biocompatible material which can be woven into gauze sheets is amenable to use. The metallic gauzes used possess the advantage of being easily cleaned and reused for many cycles of cell growth, as described below.

From this point on only 316 stainless steel gauze with a weave of 107 wires x 59 wires per inch, and a wire diameter of 160 μm (resulting in a wire spacing of 77 μm x 270 μm) will be cited. The gauze static mixer elements discussed below are constructed of this material. In addition to providing increased surface area, the gauze allowed cells to grow in multilayers. A micrograph of multilayered cells grown on this gauze is shown in Figure 3.

It was of interest to determine if the cells throughout the layers are all viable or if the cells in the center of the multilayer are necrotic. Fluorescein diacetate and ethidium bromide were used for this purpose as detailed in the Materials and Methods section (Example 3). By scanning cross-sections of the gauze using a laser confocal microscope, it was evident that the vast majority of the cells were viable based on the facts that ethidium bromide did not intercalate the nuclear DNA and that fluorescein diacetate was cleaved to yield free fluorescein. Figure 4A shows the fluorescein signal for a series of cross-sections through the cell layers. Figure 4B shows the ethidium bromide signal for the same sections. Separate coupons of cells made non-viable by treatment with methanol served as a positive control for the ethidium bromide and a

negative control for fluorescein diacetate in these experiments. This control is shown in Figure 5.

EXAMPLE 6 Planting cells on gauze static mixer elements.

A uniform plant is important for optimal culture performance and for maintaining a low, uniform cell doubling level for biologicals manufacturing. Cylindrial gauze static mixer elements composed of the same material investigated above were obtained and inserted into the bore of a sterile column, connected to a perfusion pump, a nutrient resevoir, and gas exchanger. The cylindrical reactor was oriented with the axis vertical. Although early studies were conducted with the reactor axis horizontal to promote cell sedimentation onto the growth surface, it was found that attachment to a vertical surface is as efficient, and is the preferred mode of operation. Throughout this work cells were planted at approximately 10,000 cm^. Cells have been planted in multiple element reactores where the mixing axes of the individual elements are anywhere between parallel or ninety degrees to each other. For purposes of fluid mixing and optimal reactor operation, the perpendicular configuration is preferred.

Cells did not plant in early studies employing a recirculation rate which gave 5 cm/min superficial velocity. Planting efficiency was increased by filling the reactor and allowing the cells to plant by gravity sedimentation for 2 hours. This routinely provided a planting efficiency of greater than 90% (planting efficiency = number of cells planted/total number of cells in inoculum x 100%), however the plant was nonuniform. As shown in Figure 6, the bottom element contained the most cells while the top element contained the fewest. From these data, an average settling velocity of the MRC-5 cell preparation was calculated to be 6 cm/h. In a further experiment using four elements, medium was recirculated for planting at a rate approximately equal to this settling velocity. A uniform plant was reproducibly obtained with a planting efficiency of greater than 90% for this condition (Figure 6). The above results demonstrated that medium superficial velocity

through the reactor influences both the rate of attachment of the cells and the ultimate distribution of cells in the reactor.

EXAMPLE 7 A comparison of cell growth on solid sheet and gauze static mixer elements.

Two reactors were set up and run simultaneously to delineate the differences between solid sheet and gauze elements. The solid elements were constructed of titanium, and the gauze elements were constructed of stainless steel. The surface areas were estimated to be 1300 cm^ for the solid elements and 2300 cm^ for the gauze, resulting in a gauze to solid surface area ratio of 1.8. The solid sheet elements operating in conjunction with the gauze elements provided the control for all experiments. Reactors were planted with approximately 10,000 - 20,000 cells/cm^. Reactors were run in a batch mode until the glucose concentration fell below 120 mg/dL, at which time medium perfusion was started. Both reactors were inoculated with the same harvest of MRC-5 cells, and were fed from a common reservoir of medium to eliminate any differences in these variables.

The glucose uptake rate (GUR) for each reactor is shown in Figure 7. Figure 7 A compares the reactors on a cm^ basis using the areas referred to above; these surface areas are considered approximate. Figure 7B compares the systems on a reactor medium volume basis since each reactor was the same size, i.e., five 2"x2" elements with a 1 L satellite medium vessel. Figure 7B does not embody any approximations regarding surface area. Glucose uptake rate is employed as a measure of cell growth and was used to estimate the onset of stationary phase. The last point in Figure 7 indicates the time the reactors were harvested by triton lysis. Cells were visually confluent on the solid matrix and interstices on the mesh were almost filled on the gauze elements, suggesting that more growth was possible. At this point, the medium was perfused at a rate of 0.19 L/h for the solid reactor, and the glucose concentration in the medium at this point was 1.16 g/L. The perfusion rate was 0.26 L/h for the gauze reactor, and

the medium glucose concentration was 0.69 g/L. These results demonstrate the superior metabolic activity that is obtained for the gauze reactor versus the solid reactor for the cell growth phase. The reactors were drained and the cells were lysed using a detergent buffer to extract cellular protein, a measure of the cell mass in the reactor. For these experiments, the lysis procedure to harvest the cell-associated protein from each reactor was identical: 2 L of room temperature PBS was circulated through the reactors at 100 mL/min for 5 minutes and then drained. This wash to remove medium proteins was repeated twice. Next, 1 L of the lysis buffer in the Materials and Methods section was recirculated through the reactor at 400 - 500 mL/min, 37° C, for 1 hr. This was repeated once (lysates 1 and 2), and subsequent lysates circulated longer - 2 hr for lysate 3, and 4 hr for lysate 4. The percentage of protein removed in each lysate is presented in Figure 8. When total protein calculations are converted to a cell density by the correlations presented in the Example 7 Appendix, the solid static mixer elements contained 3x10^ cells/cm^ while the gauze elements contained 6x10-* cells/cm^. The former number is believed to be representative for the solid sheet reactor since all of the cells were removed during lysis as demonstrated by swabbing the elements and viewing the swab under the microscope. However, all of the cell debris was not removed from the gauze elements as observed by microscopy, so the latter number is predictably low. On a reactor basis, the gauze elements supported greater than 3x the cell concentration of the solid elements. Provided surface area calculations are accurate, this difference in total protein is in excess of the ratio of the surface areas and is evidence of multilayer cell growth.

EXAMPLE 7- APPENDIX

SUMMARY

Total protein from MRC-5 cell lysates was quantified and compared with direct cell counts (by hemacytometer and the use of a Coulter Counter). Healthy cells in the early exponential, late exponential, and stationary growth phases as well as cells at 1, 2, and 3 weeks post infection demonstrated that total protein measurements from each culture lysate could be directly related to cell density by use of a single plot. This simple correlation provides a means to estimate the cell density in reactors that are not amenable to direct counting techniques.

Protein assay and sample preparation.

The Bio-Rad protein assay (Cat. #500-0001) is a rapid, sensitive assay used to quantify total protein. Early experiments demonstrated that the current lysis buffer (0.1% Triton X-100, 10 mM Tris-HCl, 0.1 mM MgC12) does not interfere with the Coomassie Blue based assay. For consistency all dilutions, including those of the standard protein (bovine serum albumin), are made in triton lysis buffer. The fact that triton cell lysates are often flocculent in nature indicates that samples may require further treatment prior to being assayed. Large floes lead to high protein titers. However, after two cycles of rapid freeze (in -70° C ethanol) and thaw, the floe size was reduced to the degree that a homogeneous suspension was obtained. Furthermore, by assaying several dilutions of a single sample, extraneous points that may result from a poorly prepared sample are easily detected. Samples were also centrifuged and the clarified supernatant used; this treatment provided similar results but was deemed unnecessary. Data presented here are for cell lysates frozen a minimum of two times and well vortexed prior to being assayed.

Protein concentration as a function of cell density.

For a single standard curve to allow conversion of total protein to a specific cell number, cells throughout the process must be comprised of an approximately equal quantity of protein. To test this, T-flasks were inoculated with an equal number of cells. At specified intervals in the process, two flasks were sacrificed. One flask was trypsinized and cells counted by hemacytometer and a Coulter Counter while the other flask was subjected to detergent lysis, i.e., washed twice with 0.3 mL/cm^ PBS, lysed twice with 0.08 mL/cm^ lysis buffer. This was done for cultures in the early exponential growth phase, late exponential growth phase, and stationary phase, as well as for cultures at 1 , 2, and 3 weeks post infection. The cumulative data for these experiments is presented in Figure 12. The best fit line is described by the equation: Y(10 ⁴ ) = -0.09 + 0.42X. This figure demonstrates that a single curve provides a means to estimate cell number by total protein measurements, regardless of the "state" of the cells.

The cells in this experiment were infected with HAV CR326F P28 at an MOI of 1. At 3 weeks post infection, trypsinization proved ineffective in harvesting cells. However, since the cell densities from 1 and 2 weeks post infection were approximately equal, and visible cell lysis at 3 weeks did not appear to exist, cell densities used for 3 weeks were based on data obtained at 2 weeks. Figure 12 is replotted, indicating the time of each specific data point. By discounting the data at 3 weeks post infection, the slope and intercept of the curve is not greatly affected (see Figure 13).

Using the above correlation, total protein data for several culture lysates yielded an average of 490 μg/mL protein. This translates to approximately 3.3 x 10^ cells/cm^, and is in general agreement with cell densities determined using a hemacytometer on control T flasks. In summary, it is believed that total protein of Triton cell lysates provides a useful approximation of the cell density at the time of virus harvest.

EXAMPLE 8 Production of HAVAg on gauze static mixer elements.

HAVAg production was demonstrated for a reactor employing gauze static mixer elements. The objective of this experiment was to determine if viral propagation occurred on stainless steel gauze elements supporting multilayer growth; it was not an optimized system for the production of HAVAg. For this run, complete medium replacements were employed rather than medium perfusion. A GUR profile for this run is shown in Figure 9; the GUR's calculated for this figure are based on glucose values obtained the day after medium was refed. Cells were infected with an MOI of about 1 on day 7. Cell density for MOI calculations was estimated based on the glucose consumption rate at the time of infection. As observed with processes such as where cells are grown on solid surfaces, the glucose uptake rate plateaus shortly after infection and is followed by a decline, consistent with cell infection. The observed decline in Figure 9 is precipitous and was likely due to a uniform degree of infection . The viral antigen was harvested 21 days after infection by Triton detergent lysis. The protocol for lysing the cells is outlined below.

Lysate 1 : recirculate lysis buffer through static mixer for 1 hr at 37° C. Lysate 2: recirculate lysis buffer through static mixer for 1 hr at 37° C. Lysate 3: sonicate for 1 hr in Branson sonicator bath containing 0.7 L lysis buffer.

Lysate 4: freeze elements in 1 L lysis buffer for 12 hr, thaw, and sonicate for lhr.

The percent of total protein and HAVAg released in these washes is indicated in Figure 10. The first 2 lysates employing recirculation were very effecient in removing HAVAg from the cells. This experiment employed only 2 PBS washes before lysis, and was not completely efficient. Therefore the total protein percent of lysate 1 is artificially high and that of 2, 3, and 4 are low. A single pooled sample was assayed for HAVAg by EIA prior to freezing. Based on the percentages

of each lysate determined later, this correlated to an antigen yield per surface area twice that achieved on solid surface reactors. The objective of this experiment was realized in that viral propagation as indicated by HAVAg yield occurred. This experiment demonstrates the utility of the gauze static mixer reactor for production of HAV.

Although the reactor was likely harvested late based on the GUR profile, the antigen yields were relatively high, indicating susceptibility of multilayers to infection. The data also indicate the importance of MOI in HAV production; by infecting at a MOI of 1 , it appears that the process duration should have been decreased by more than one week and harvested on day 20. Based on a variety of data, the time of infection and MOI are important variables for a manufacturing process; these parameters are important in decreasing the length of the culture. Cell growth experiments indicate that infected cells continue to grow at a rate similar to uninfected cells, even for cultures infected at high MOI with CR326F P28 (Figure 11). This is despite infection and a temperature shift to 32° C, the growth permissive temperature for this strain of virus. The maximum specific growth rate calculated from these curves is 0.015 h ^" approximately half the growth rate of uninfected cells at 37° C. The fact that the cells behave similarly during the first week of HAV infection is important in the operation of a reactor, in that one would like to realize the highest cell concentration possible regardless of infection time and MOI. This experiment demonstrates that optimization of cell growth duration, infection duration, and MOI are intertwined.

EXAMPLE 9 Cleaning the static mixer elements.

It is desirable to possess a reusable growth surface for biologicals manufacture to eliminate waste disposal, enable automation of the reactor process, and enhance culture reproducibility. Currently, the gauze elements are rinsed with distilled water to remove medium salts (500 mL/min for 1 h), placed in 1 L of a 5% (w/v) NaOH solution and autoclaved for 45 min. After cooling, elements are rinsed with distilled

water until free of NaOH (as determined by rinsate pH). This procedure is extremely effective in removing cell debris by microscopic inspection. It is speculated that a much less harsh treatment would result in equivalent cleaning, and be more appropriate for a clean-in- place manufacturing reactor. Gauze elements treated by this process have been reused for more than ten cycles with no change in cell growth properties. Elements in the above vims growth study were prepared in this manner, demonstrating the utility of used, cleaned elements for HAV propagation.