Case 1:18-cv-00924-CFC Document 399-4 Filed 10/07/19 Page 1 of 17 PageID #: 30637
`Case 1:18-cv-00924-CFC Document 399-4 Filed 10/07/19 Page 1 of 17 PageID #: 30637
`
`(cid:40)(cid:59)(cid:43)(cid:44)(cid:37)(cid:44)(cid:55)(cid:3)(cid:23)(cid:3)
`
`EXHIBIT 4
`
`(cid:3)
`
`

`

`Case 1:18-cv-00924-CFC Document 399-4 Filed 10/07/19 Page 2 of 17 PageID #: 30638
`
`Appl Microbiol Biotechnol (2005) 68: 283–291
`DOI 10.1007/s00253-005-1980-8
`
`MINI-REVIEW
`
`Michael Butler
`Animal cell cultures: recent achievements and perspectives
`in the production of biopharmaceuticals
`
`Received: 11 February 2005 / Revised: 23 March 2005 / Accepted: 31 March 2005 / Published online: 16 April 2005
`# Springer-Verlag 2005
`
`Abstract There has been a rapid increase in the number
`and demand for approved biopharmaceuticals produced
`from animal cell culture processes over the last few years.
`In part, this has been due to the efficacy of several hu-
`manized monoclonal antibodies that are required at large
`doses for therapeutic use. There have also been several
`identifiable advances in animal cell technology that has
`enabled efficient biomanufacture of these products. Gene
`vector systems allow high specific protein expression and
`some minimize the undesirable process of gene silencing
`that may occur in prolonged culture. Characterization of
`cellular metabolism and physiology has enabled the design
`of fed-batch and perfusion bioreactor processes that has
`allowed a significant improvement in product yield, some
`of which are now approaching 5 g/L. Many of these pro-
`cesses are now being designed in serum-free and animal-
`component-free media to ensure that products are not
`contaminated with the adventitious agents found in bovine
`serum. There are several areas that can be identified that
`could lead to further improvement in cell culture systems.
`This includes the down-regulation of apoptosis to enable
`prolonged cell survival under potentially adverse condi-
`tions. The characterization of the critical parameters of
`glycosylation should enable process control to reduce the
`heterogeneity of glycoforms so that production processes
`are consistent. Further improvement may also be made by
`the identification of glycoforms with enhanced biological
`activity to enhance clinical efficacy. The ability to produce
`the ever-increasing number of biopharmaceuticals by ani-
`mal cell culture is dependent on sufficient bioreactor capacity
`in the industry. A recent shortfall in available worldwide
`culture capacity has encouraged commercial activity in
`contract manufacturing operations. However, some ana-
`
`M. Butler (*)
`Department of Microbiology,
`University of Manitoba,
`Buller Building,
`Winnipeg, Manitoba, Canada, R3T 2N2
`e-mail: butler@cc.umanitoba.ca
`Tel.: +1-204-4746543
`Fax: +1-204-4747603
`
`lysts indicate that this still may not be enough and that future
`manufacturing demand may exceed production capacity as
`the number of approved biotherapeutics increases.
`
`Introduction
`
`Although animal cell cultures have been important at a
`laboratory scale for most of the last 100 years, it was the
`initial need for human viral vaccines in the 1950s (par-
`ticularly for poliomyelitis) that accelerated the design of
`large-scale bioprocesses for mammalian cells (Kretzmer
`2002). These processes required the use of anchorage-de-
`pendent cells and the modern version of this viral vaccine
`technology currently employs microcarrier support sys-
`tems that can be used in pseudosuspension cultures de-
`signed in stirred tank bioreactors.
`However, more recently the enhanced interest in mam-
`malian cell culture bioprocesses is associated with re-
`combinant protein technology developed in the 1970s and
`1980s. The first human therapeutic protein to be licensed
`from this technology in 1982 was recombinant insulin (Hu-
`mulin from Genentech) but the relative structural simplicity
`of this molecule allowed its large-scale production to be
`developed in Escherichia coli, which is fast growing and
`robust compared to mammalian cells. It was soon realised
`that the subsequent targets for recombinant therapeutics
`were more complex and required the post-translational
`metabolic machinery only available in eukaryotic cells. At
`the present time there are up to 30 licensed biopharmaceu-
`ticals produced from mammalian cell bioprocesses (Walsh
`2003; Pavlou 2003, Molowa and Mazanet 2003). These are
`defined as recombinant proteins, monoclonal antibodies
`and nucleic acid-based products. Since 1996, the chimeric
`and humanized monoclonal antibodies have dominated this
`group with such blockbuster products as Rituxan, Remi-
`cade, Synagis and Herceptin (Brekke and Sandie 2003;
`Pavlou 2004). A chimeric antibody (e.g. Rituxan) consists
`of a molecular construct in which the mouse variable region
`is linked to the human constant region. A further step to
`humanizing an antibody can be made by replacement of the
`APPX 0220
`
`

`

`Case 1:18-cv-00924-CFC Document 399-4 Filed 10/07/19 Page 3 of 17 PageID #: 30639
`
`284
`
`murine framework region, leaving only the complementar-
`ity determining regions (CDRs) that are of murine origin.
`These hybrid construct molecules are far less immunogenic
`than their murine counterparts and have serum half-lives of
`up to 20 days.
`In this review, key factors for the recent achievements
`in the production of biopharmaceuticals from animal cell
`culture processes are discussed.
`
`Cell line transfection and selection
`
`The ability to produce and select for a high-producing
`animal cell line is key to the initial stages of the devel-
`opment of a cell culture bioprocess (Wurm 2004; Andersen
`and Krummen 2002). Chinese hamster ovary (CHO) cells
`have become the standard mammalian host cells used in the
`production of recombinant proteins, although the mouse
`myeloma (NS0), baby hamster kidney (BHK), human em-
`bryonic kidney (HEK-293) or human-retina-derived (PER-
`C6) cells are alternatives. All these cell lines have been
`adapted to grow in suspension culture and are well suited
`for scale-up in stirred tank bioreactors. The advantage of
`CHO and NSO cells is that there are well-characterised
`platform technologies that allow for transfection, amplifi-
`cation and selection of high-producer clones. Transfection
`of cells with the target gene along with an amplifiable gene
`such as dihydrofolate reductase (DHFR) or glutamine syn-
`thetase (GS) has offered effective platforms for expression
`of the required proteins. In these systems, selective pressure
`is applied to the cell culture with an inhibitor of the DHFR
`or GS enzymes that causes an increase in the number of
`copies of the transfected genes including the target gene.
`The DHFR system is routinely used with CHO cells
`−
`). The target gene
`deficient in the DHFR activity (DHFR
`is delivered to the cells along with the DHFR marker gene,
`usually on the same plasmid vector (Gasser et al. 1982;
`Lucas et al. 1996). The expression vector normally con-
`tains a strong viral promoter to drive transcription of the
`recombinant gene and this is delivered into the cells by
`one of a number of possible non-viral transfer techniques.
`These include calcium phosphate, electroporation, lipofec-
`tion or polymer-mediated gene transfer. The transfected
`cells are selected in media requiring the activity of DHFR
`for nucleotide synthesis and cell growth. Exposure of the
`cells to several rounds of gradually increased concentra-
`tions of the DHFR enzyme inhibitor, methotrexate, pro-
`motes amplification of the DHFR and the co-transfected
`target gene. Methotrexate treatment enhances specific pro-
`tein production following an increased gene copy number,
`which can be up to several hundred in selected cells.
`The glutamine synthetase expression system is an alter-
`native that works as a dominant selectable marker, which
`is an advantage because this does not require the use of
`specific mutant cells (Bebbington et al. 1992). CHO cells
`that contain endogenous enzyme activity can be used al-
`though NS0 cells are preferred because of the absence of
`GS. This means that a lower activity of an enzyme inhib-
`
`itor can be used for selection and amplification in NS0. GS
`enzyme allows the synthesis of glutamine intracellularly
`and so the transfected cells are selected in a glutamine-free
`media. The added advantage of this is that the cell cultures
`produce less ammonia, which is a potentially toxic met-
`abolic by-product of mammalian cells that affects protein
`glycosylation and may inhibit cell growth. Gene amplifi-
`cation in this system is mediated by methionine sulphox-
`imine (MSX), which is required at concentrations of 10–
`100 μM to provide clones with amplified genes and suf-
`ficiently high specific productivities. Typically copy num-
`bers of only four to ten genes per cell are found in these
`cells but they give as high expression levels as the cells
`from the DHFR system. The advantage of this is that the
`GS high-producer clones can be produced in around 3
`months, which is half the time it takes for the selection of
`DHFR clones.
`High yields of recombinant proteins can also be pro-
`duced from a human cell line, notably PER.C6, which was
`created by immortalizing healthy human embryonic retina
`cells with the E1 gene of adenovirus (Jones et al. 2003).
`This cell line has been well characterized and has been
`shown to be able to produce high levels of recombinant
`protein with relatively low gene copy numbers and with-
`out the need for amplification protocols. The added value
`of these cells is that they ensure the recombinant proteins
`produced receive a human profile of glycosylation.
`Screening for a high-producer clone can be a lengthy
`process that depends upon assaying the secreted proteins
`to determine productivities of all candidate clones. High-
`throughput selection systems have been devised based on
`rapid assays or the use of flow cytometry to identify clones
`that have an appropriate product marker on the cell surface
`(Borth et al. 2001; Carroll and Al-Rubeai 2004). A cell clone
`−1 day
`−1 can
`with a specific productivity of up to 10 pg cell
`be produced fairly routinely for recombinant protein pro-
`duction. However, higher specific productivity (up to 90 pg
`
`−1 day−1) may be possible with improvements in vector
`cell
`technology and further understanding of the parameters
`that control protein expression in the cell (Wurm 2004).
`
`Stability of gene expression
`
`The stability of selected clones over long-term culture is a
`critical parameter for commercial production (Kim et al.
`1998). The application of selective pressure such as meth-
`otrexate in the case of the DHFR selection system causes
`gene amplification but a proportion of these genes are
`unstable and removal of the selective agent, as is necessary
`in production cultures, results in a gradual loss of the gene
`copy number. Fann et al. (2000) reported the stepwise ad-
`aptation of tissue plasminogen activator-producing CHO
`cell lines to 5 μM of methotrexate, which resulted in a
`maximum specific recombinant protein production of 43 pg
`
`−1 day−1, but on removal of the methotrexate the max-
`cell
`
`−1 day−1 within
`imum productivity decreased to 12 pg cell
`40 days. This decrease in productivity could be correlated
`
`APPX 0221
`
`

`

`Case 1:18-cv-00924-CFC Document 399-4 Filed 10/07/19 Page 4 of 17 PageID #: 30640
`
`with a reduction of gene copy number for individual clones.
`Barnes et al. (2004) studied the stability of antibody ex-
`pression from NS0 cells amplified with the GS system.
`They reported that there was a loss of mRNA for the re-
`combinant protein over long-term culture but this was only
`reflected in a decrease in protein expression if the mRNA
`was below a threshold level. This indicates that selection for
`clones for high levels of recombinant mRNA may be useful
`as a predictor of stable protein production. Above a sat-
`uration level of mRNA it is argued that the limitation to
`protein expression resides in the translational/ secretory
`machinery of the cell.
`Production of a high-producer clone is dependent on the
`integration of the expression vector in the host cell genome.
`The site of integration has a major effect on the transcription
`rate of the recombinant gene and progressive gene silencing
`can occur over successive culture passages after clonal
`isolation. This is thought to be associated with the spread of
`heterochromatin structure, which is condensed and tran-
`scriptionally silent. Most expression systems cause random
`transgene integration into the host cell and this leads to
`positional effects that cause variable expression and stabil-
`ity. However, the genetic control elements that are respon-
`sible for establishing a transcriptionally active transgene are
`not fully understood. Vectors in which the genes are flanked
`with insulators, boundary elements or ubiquitous chromatin
`opening elements may promote stable expression by insu-
`lating the transgene from positional effects of the chroma-
`tin. These elements that can be incorporated into expression
`vectors include matrix (or scaffold) attachment regions
`(MAR or SAR) that allow open chromatin structure to be
`maintained. This can allow higher efficiency of expression
`of the integrated genes (Kim et al. 2004). Gene regulatory
`elements associated with ubiquitously expressed house-
`keeping genes have been recently isolated (Antoniou et al.
`2003). These regulatory elements appear to confer a dom-
`inant chromatin opening function and give rise to an ability
`to resist transgene silencing. These ubiquitous chromatin
`opening elements (UCOE) have also been incorporated into
`transgene vectors to prevent gene silencing and give con-
`sistent, stable and high-level gene expression irrespective
`of the chromosomal integration site (Haines et al. 2004).
`
`Culture modes
`
`A producer cell clone may be grown in batch cultures to
`above 106 cells/mL over 3–4 days to allow synthesis and
`product secretion. The limits for growth and production
`are related to the accumulation of metabolic by-products,
`such as ammonia and lactate, or the depletion of nutrients,
`such as glucose or glutamine (Butler and Jenkins 1989).
`The growth of cells can be extended if these limitations are
`addressed through perfusion culture in which the constant
`supply of nutrients and the removal of media can lead to
`cell densities of at least 107 cells/mL (Butler et al. 1983).
`These principles have been extended to fed-batch cultures,
`which have been shown to be operationally simple, reli-
`able and flexible for multi-purpose facilities (Bibila and
`
`285
`
`Robinson 1995, Cruz et al. 2000; Xie and Wang 1997). The
`most successful strategies involve feeding concentrates of
`nutrients based upon the predicted requirements of the cells
`for growth and production. This can involve slow feeding of
`low concentrations of key nutrients. The maintenance of
`low concentration set points of the major carbon substrates
`enables a more efficient primary metabolism with leads to
`lower rates of production of metabolic by-products, such as
`ammonia and lactate. As a result the cells remain in a pro-
`ductive state over extended time frames. The strategic use of
`fed-batch cultures has enabled considerable enhancement
`of yields from these processes. This is often combined with
`a biphasic strategy of production in which cell proliferation
`is allowed in the first phase so that high cell densities ac-
`cumulate, followed by a phase in which cell division is
`arrested to allow cells to attain a high specific productivity.
`In this type of strategy, growth can be arrested by a decrease
`in culture temperature (Fox et al. 2004; Yoon et al. 2003).
`By directly supplying cells with a balanced nutrient feed, a
`fed-batch culture can now be expected to yield upwards of
`2 g/L of recombinant protein, which is probably at least
`tenfold higher than the maximum that could be expected by
`a simple batch culture in standard culture medium.
`Animal cell cultures are normally grown in stainless
`steel, stirred tank bioreactors that are designed with im-
`pellers that minimize shear forces (Kretzmer 2002). Pro-
`ducer cells can be made to be sufficiently robust in this
`environment if they are provided with suitable growth
`media and gas sparging is carefully controlled. The capacity
`of commercial bioreactors for animal cells has gradually
`increased over the past two decades, with capacities now
`reported up to 20,000 L from some of the larger biopharma-
`ceutical companies. Airlift bioreactors have also been ap-
`plied to large-scale animal cells and these have been shown
`to be efficient for protein production.
`Perfusion cultures are more demanding to set up at a
`large scale but they have the potential advantage of allow-
`ing a continuous stream of product over several weeks or
`even months (Mercille et al. 2000). A further advantage is
`the rapid removal of any potentially labile products from
`the culture environment. An effective cell separator will
`allow the protein-containing media to be fed directly into a
`chromatography column suitable for extraction and down-
`stream processing (Shirgaonkar et al. 2004; Castilho and
`Medronho 2002; Wen et al. 2000). A further advantage of
`this mode of culture is that the bioreactor may be up to ten
`times smaller for the production of the same quantity of
`product (Ryll et al. 2000).
`This area of bioprocess design will become of even
`greater importance as some of the first-generation block-
`buster drugs (e.g. erythropoietin, human growth hormone
`and α-interferon) start being produced as generics (Walsh
`2003). Eleven biopharmaceuticals with combined annual
`sales of $13.5 billion lose patent protection in 2006 (Walsh
`2003). The challenge then will be to produce bioequiva-
`lents in efficient low-cost bioprocesses.
`There are several challenging areas of bioprocess de-
`velopment that are required to be addressed to ensure the
`future success of animal cell culture processes. These in-
`
`APPX 0222
`
`

`

`Case 1:18-cv-00924-CFC Document 399-4 Filed 10/07/19 Page 5 of 17 PageID #: 30641
`
`286
`
`clude serum-free media, apoptosis, glycomics and the ca-
`pacity crunch.
`
`Apoptosis
`
`Serum-free media
`
`Bovine serum was used as a supplement of cell culture
`media for several decades. It is a rich source of hormones,
`growth factors and trace elements that promote rapid cell
`growth and also its high albumin content ensures that the
`cells are well protected from potentially adverse condi-
`tions such as pH fluctuations or shear forces. However, the
`composition of serum is variable and undefined, which
`leads to inconsistent growth and productivity. Early at-
`tempts to develop serum-free substitutes incorporated such
`components as insulin, transferrin, albumin and cholesterol.
`However, the mad cow crisis in the beef industry alerted a
`concern for the use of animal serum and any other animal-
`derived components in the production of biotherapeutics.
`This has now led to a strong demand for cell culture for-
`mulations that are free of all animal components. The chal-
`lenge that this demand poses is to be able to identify effective
`substitutes for all the growth-promoting factors that are
`present in serum. It turns out that producer cell lines are quite
`fastidious in their growth requirements and that such re-
`quirements vary considerably from one cell line to another.
`Therefore, it has not been possible to design a single serum-
`free formulation to act as a serum substitute suitable for the
`growth of all cell lines. In fact even different clones of CHO
`cells may require different formulations for optimal growth.
`This has given rise to a strong drive for the development of
`serum-free and animal-component-free formulations that are
`tailored to the needs of specific producer cell lines.
`There are several strategies that can be used to design
`these formulations. Combinations of standard basal media
`may be tested to determine those that result in good cell
`growth and productivity at minimal serum levels. In some
`cases metabolic analysis may help in media design. For
`example, NS0 myeloma cells lack a functional pathway
`for cholesterol synthesis and so cholesterol is required as a
`lipoprotein supplement in the medium (Gorfien et al. 2000).
`Protein hydrolysates from non-animal sources have been
`found to provide good growth promotion in some culture
`systems (Sung et al. 2004). Analysis of the depletion of
`media components may lead to the identification of specific
`nutrients that may be required at higher supplement levels
`or for inclusion in feeding regimes. Statistical approaches
`can be used, such as the Plackett–Burman experimental
`design, so that mixtures of components may be tested si-
`multaneously in matrix experiments of growth and produc-
`tivity (Castro et al. 1992). Another original approach is the
`identification by microarray analysis of specific receptors
`expressed during cell growth, so that corresponding ligands
`may be incorporated into the medium (Donahue 2004).
`These approaches are presently being used by various
`specialized media companies to customise animal-compo-
`nent-free formulations for the production of the plethora of
`recombinant proteins that are being introduced into the
`market.
`
`Cell cultures are often terminated because of cell death
`that may be caused by one of several factors including nu-
`trient depletion, metabolic by-product accumulation, exces-
`sive shear forces or hypoxia. Cell death may be by necrosis
`caused by extreme conditions resulting in physical damage
`to the cells. Alternatively and more commonly, cell death in
`a bioreactor occurs by apoptosis, which is a form of pro-
`grammed cell death regulated through a cellular cascade of
`activities in response to one of the factors mentioned above
`(Arden and Betenbaugh 2004). Characteristic changes in-
`clude chromatin shrinkage followed by membrane bleb-
`bing and the formation of apoptotic bodies. The DNA of
`the cell is fragmented and this can be the basis of an assay
`to quantify apoptosis in a cell population.
`It is of considerable value to be able to prevent or inhibit
`apoptosis in culture in order to extend the time of high cell
`viability and prolong protein production. There are two
`strategies that can be used for this. The cellular environment
`can be manipulated through media supplementation or
`the intracellular environment can be modified by genetic
`engineering.
`Nutrient feeding can provide protection and this is nor-
`mally used as the first preventive measure to control the
`cellular environment to delay apoptosis. Serum is known to
`contain unidentified anti-apoptotic factors that can offer
`protection (Zanghi et al. 1999). However, the serum-free
`formulations that are required for production processes
`make the cells more vulnerable to apoptosis. Some supple-
`ments such as suramin (Zanghi et al. 2000) or insulin growth
`factor (Sunstrom et al. 2000) may provide independent anti-
`apoptotic protection in serum-free cultures. There are also
`other specific caspase inhibitors available to suppress apo-
`ptosis (Tinto et al. 2002) but their expense in large-scale
`cultures is likely to be prohibitive.
`Genetic strategies involve the transfection of anti-apo-
`ptotic genes such as bcl-2 or bcl-x2 into a host cell. The
`expression of the corresponding proteins inhibits the release
`of pro-apoptotic molecules from the mitochondria and may
`prolong the viability of the cell. This strategy has been
`shown to work for several cell lines, which have shown
`higher viabilities and improved robustness under conditions
`that would normally be expected to cause apoptosis (Tey
`et al. 2000; Mastrangelo et al. 2000; Kim and Lee 2002).
`
`Glycosylation
`
`Animal cells are used for biomanufacture because of their
`capabilities of adding carbohydrates (glycans) to synthe-
`sised proteins (Butler 2004). These are produced as pools of
`different glycoforms with varying glycan structures at-
`tached to a single peptide backbone with a known amino
`acid sequence. The basic protein structures can be controlled
`and directed by the expression of appropriate genetic se-
`quences. However, controlling the pool of glycan structures
`(glycomics) that occupy a recombinant protein is still dif-
`ficult. Variations may be found in the site occupancy (mac-
`
`APPX 0223
`
`

`

`Case 1:18-cv-00924-CFC Document 399-4 Filed 10/07/19 Page 6 of 17 PageID #: 30642
`
`roheterogeneity) or in the structure of attached glycans
`(microheterogeneity).
`For the production of a recombinant protein as a bio-
`therapeutic it is essential to ensure that a consistent glyco-
`sylation profile is maintained between batches (Restelli
`and Butler 2002). However, this may not be so easy to con-
`trol given that the extent of glycosylation may decrease over
`time in a batch culture (Curling et al. 1990). This is likely to
`be due to the depletion of nutrients, particularly glucose or
`glutamine, which have been shown to limit the glycosyla-
`tion process (Hayter et al. 1992; Nyberg et al. 1999). Fed-
`batch strategies should also be designed to ensure that the
`concentrations of these key nutrients do not decrease to
`a critical level that could compromise protein glycosyla-
`tion (Xie and Wang 1997). These lower levels were found to
`be <0.1 mM glutamine and <0.7 mM glucose for the pro-
`duction of γ-interferon from CHO cells (Chee et al. 2005).
`In one report it is suggested that site occupancy could vary
`with the growth state of cells and correlates with the frac-
`tion of cells in the G0/G1 phase of the cell cycle (Andersen
`et al. 2000). This suggests a mechanism by which glyco-
`sylation efficiency improves at a reduced rate of protein
`translation.
`Culture parameters such as nutrient content, pH, tem-
`perature, oxygen or ammonia, may have a significant effect
`on the distribution of glycan structures found on the re-
`sulting recombinant protein (microheterogeneity). This of
`course is of major concern in trying to produce consistent
`biopharmaceuticals. Decreased sialylation or altered pat-
`terns of glycan branching occur when the ammonia level
`accumulates in culture (Andersen and Goochee 1994;
`Zanghi et al. 1998; Yang and Butler 2000). Non-optimal pH
`conditions (<6.9 and >8.2) have also been shown to alter the
`pattern of glycosylation (Rothman et al. 1989; Borys et al.
`1993). Reduced terminal galactosylation has been shown in
`the glycans of immunoglobulin (IgG) produced under low
`oxygen conditions (Kunkel et al. 1998). Nabi and Dennis
`(1998) observed an increase in the polylactosamine content
`of a protein produced at lower temperatures and attributed
`this to changes in the transit time through the Golgi.
`The pattern of protein glycosylation is dependent on the
`expression of various glycosyltransferase enzymes that are
`present in the Golgi of the cell. Differences in the relative
`activity of these enzymes among species can account for
`significant variations in structure. In one systematic study
`of glycan structures of IgG produced from cells of 13
`different species significant variation was found in the
`proportion of terminal galactose, core fucose and bisecting
`GlcNAc (Raju et al. 2000). The structure of sialic acid
`may also vary, with N-glycolyl-neuraminic acid (NGNA)
`found in goat, sheep and cows rather than the N-acetyl-
`neuraminic acid (NANA) found in humans. NGNA is the
`predominant sialic acid in mice, but CHO-produced gly-
`coproteins have predominantly NANA, although a small
`proportion (up to 15%) of NGNA can occur (Baker et al.
`2001). These differences in glycan structure are important
`tential immunogenicity of these structures in humans. Mouse
`cells express the enzyme α1,3 galactosyltransferase, which
`
`287
`generates Galα1,3-Galβ1,4-GlcNAc residues that are high-
`ly immunogenic in humans (Jenkins et al. 1996). Fortu-
`nately, this enzyme appears to be inactive in CHO and BHK
`cells, which are the most commonly used cell lines for the
`production of recombinant proteins. However, both CHO
`and BHK show differences in their potential for glycosy-
`lation compared to human cells. The sialyl transferase en-
`zyme (α2,6 ST) that normally provides an α2,6 linkage for
`terminal sialic acid in glycoproteins produced in humans is
`absent in the hamster and thus CHO and BHK cells pro-
`duce exclusively α2,3 terminal sialic acid residues. Further-
`more, the absence of a functional α1,3 fucosyltransferase
`in CHO cells prevents the addition of peripheral fucose
`residues and the absence of N-acetylglucosaminyltransfer-
`ase III (Gn TIII) prevents the addition of bisecting GlcNAc
`to glycan structures (Jenkins and Curling 1994). However,
`these differences in glycosylation potential between CHO
`and human cells do not appear to result in glycoproteins
`that are immunogenic. Natural human erythropoietin (EPO)
`consists of a mixture of sialylated forms: 60% are 2,3 linked
`and 40% are 2,6 linked. Because of the restricted sialyla-
`tion capacity of CHO cells, the commercially available
`EPO is sialylated entirely via the α2,3 linkages. Never-
`theless, recombinant EPO produced from CHO cells has
`proven to be a highly effective therapeutic agent with no
`evidence of an adverse physiological effect due to the
`structural differences in terminal sialylation.
`The production of specific protein glycoforms may allow
`the possibility of even more efficacious drugs (Shriver et al.
`2004). Functional glycomics is an expanding area of sci-
`ence that attempts to understand the physiological function
`of specific carbohydrate groups. This approach established
`the importance of the sialylation of EPO with the discovery
`that the removal of sialic acid groups from the glycans re-
`sulted in a significantly reduced half-life in the blood stream
`(Erbayraktar et al. 2003). Protein engineering has allowed
`the creation of a modified EPO with two extra glycan at-
`tachment sites and with the potential to incorporate eight
`extra sialic acid groups per molecule. This has led to a new-
`generation EPO called darbepoetin, which has a three times
`higher drug half-life (Egrie et al. 2003). This strategy of
`enhancing the half-life of a biotherapeutic has also been
`successful for other recombinant proteins such as follicle-
`stimulating hormone (Perlman et al. 2003) and thyroid-
`stimulating hormone (Thotakura et al. 1991).
`Structural changes of glycans can also be brought about
`by metabolic engineering of the host cell line. This in-
`cludes gene knockout of already expressed glycosyltrans-
`ferases or the insertion of novel activities (Weikert et al.
`1999). The presence of a bisecting N-acetylglucosamine
`(Umana et al. 1999; Davies et al. 2001) or the absence of
`fucose (Shields et al. 2002; Shinkawa et al. 2003; Okazaki
`et al. 2004) in the conserved glycan of an IgG antibody
`has been shown to enhance attachment to Fc receptors and
`result in an increase in antibody-dependent, cell-mediated
`cytotoxicity (ADCC). This has been of value in the design
`of antibody therapeutics. For example, recent work with
`Herceptin, which is a novel humanized antibody approved
`
`APPX 0224
`
`

`

`Case 1:18-cv-00924-CFC Document 399-4 Filed 10/07/19 Page 7 of 17 PageID #: 30643
`
`288
`
`for the treatment of breast cancer, has shown that a glyco-
`form with no fucose has a 53 times higher binding ca-
`pacity to an Fc receptor that triggers its therapeutic activity
`(Shinkawa et al. 2003). This enhancement of ADCC al-
`lows the antibody to be effective at lower doses. Afuco-
`sylated antibodies can be produced from cells in which the
`gene for fucosyl transferase has been removed by gene
`knockout technology.
`Complete glycosylation of recombinant proteins is usu-
`ally associated with maximisation of galactosylation and
`sialylation. Often these two processes are incomplete and
`this gives rise to considerable glycan structural variation.
`CHO cells can be engineered with a combination of human
`β1,4-galactosyltransferase and α2,3-sialyltransferase to en-
`sure high activities of these enzymes. The recombinant pro-
`teins produced by these cells exhibited greater homogeneity
`compared to controls and increased terminal sialic acid re-
`sidues (Weikert et al. 1999). An alternative approach in-
`volves glycoengineering of the proteins in vitro (Raju et al.
`2001). Preparations of these terminal transferase enzymes
`can be immobilized so that glycoproteins can be galacto-
`sylated and sialylated in the presence of appropriate ga-
`lactose and sialic acid donors.
`
`The capacity crunch
`
`With an increase in the number and demand for recombi-
`nant biopharmaceuticals, there is a requirement for greater
`biomanufacturing capacity. This created a major problem in
`2001 when the demand for Enbrel, a recombinant fusion
`protein commercialized by Immunex for the treatment of
`rheumatoid arthritis, exceeded expectations. However, there
`was insufficient large-scale culture manufacturing capacity
`to meet this clinical demand, even by contract manufactur-
`ers available