Castilho Leda R., Moraes Angela Maria (Ed.) Animal Cell Technology: From Biopharmaceuticals to Gene Therapy

Подождите немного. Документ загружается.

Golgi. However, in lower eukaryotes such as fungi and yeast the high-

mannose structures are released from the cell as end products of glycosyla-

tion (Meynial-Salles and Combes, 1996).

There is a great deal of variability of N-glycan structures attached to

any protein. This arises through macroheterogeneity, which is deﬁned as

variable occupancy of potential sites of attachment and microheterogene-

ity, which is deﬁned as variable structural forms at each site.

6.2.1.1 Macroheterogeneity

The occupation of a sequon occurs cotranslationally in the ER and

involves the transfer of a consensus oligosaccharide structure

(Glc

Man

GlcNAc

) from a dolichol-linked pyrophosphate donor to the

Asn side chain of an available sequon. Only around 65% of all appropriate

glycoprotein sequons are occupied and the oligosaccharyl transfer reaction

appears to depend upon various cellular factors including the protein

translation rate, the availability of the oligosaccharyl donor, and the

activity of the enzyme. Petrescu et al., (2004) surveyed the occupation of

sequons on several thousand proteins and found a signiﬁcantly large

number of occupied glycans that have low accessibility in the folded

protein. Although this may seem surprising, it must be realized that

protein glycosylation precedes folding and so the inaccessibility of a

sequon in the fully folded protein does not mean that it is necessarily

inaccessible to the OST enzyme. The glycans may play a part in maintain-

ing the folded structure by reducing the conformational freedom of the

local peptide backbone (Petrescu et al., 2004). However, disulﬁde bridge

Asn

Complex

Hybrid

High mannose

⫽ Fuc

⫽ GlcNAc

⫽ Man

⫽ Gal

⫽ NeuAc

Figure 6.3

N-glycan structures of glycoproteins. There are three main types of glycan

structure, which are built on the core structure. The high mannose structures are

the main type of structures found in fungi. In animal cells these are generally

intermediates that are transferred to the Golgi for further processing. The complex

glycans consist of a variety of structures added in sequence to the trimannosyl

core. The hybrid structures consist of at least one antenna from the trimannosyl

core with a complex structure and one with a high mannose structure.

132 Animal Cell Technology

formation is also co-translational in the ER and may interfere with the

accessibility of some sequons, leaving them unoccupied. For some proteins

such as tissue-type plasminogen activator (tPA) this may lead to variable

occupancy of a speciﬁc site and macroheterogeneity of glycoforms (Allen

et al., 1995).

A further consideration is the position of the sequon in relation to the

primary structure of the protein. Statistical analysis of a large number of

glycoproteins has indicated that the frequency of non-glycosylated se-

quons increases toward the C-terminus (Gavel and von Heijne, 1990). The

critical distance appears to be 60 amino acid residues from the C-terminus

when reduced glycan occupation occurs. This distance corresponds to the

distance between the ribosome P-site and the active site of the OST and it

has been hypothesized that the protein chain is not available for N-glycan

attachment once it is released from the ribosome. However, this phenom-

enon of poor glycosylation efﬁciency toward the C-terminus does not

appear to be universal for all proteins (Walmsley and Hooper, 2003).

6.2.1.2 Microheterogeneity

The transferase reactions in the Golgi do not always go to completion and

so give rise to heterogeneity of the ﬁnal glycan structure (microhetero-

geneity). This heterogeneity is evident as variable antennarity, with the

number of branches from the central mannose of the core structure

ranging typically between two (biantennary) and four (tetra-antennary).

Terminal sialylation of the antenna, fucosylation of the innermost core

GlcNAc (proximal) or the outer arm GlcNAc (peripheral), and addition

of a ‘‘bisecting’’ GlcNAc to the central core Man residue are also examples

of glycan processing that is variable and may be incomplete in the Golgi.

A diverse range of glycan forms may be produced in some recombinant

proteins. For example, the HIV envelope glycoprotein (gp120) produced

from CHO cells contains 24 occupied sequons, of which 13 are complex

glycans and 11 are high mannose or hybrid structural forms (Leonard

et al., 1990).

6.2.2 O-linked glycans

Although most attention has been focused on N-linked glycosylation of

proteins, O-linked glycans are smaller structurally but equally ubiquitous

among eukaryotic glycoproteins. The most common form of O-glycan

attached to glycoproteins from mammalian cells is the mucin-type, which

involves the addition of N-acetylgalactosamine to a serine or threonine

residue in a protein. The O-glycan is added post-translationally to the

fully folded protein. No consensus sequence has been identiﬁed, although

glycosylation often occurs in a region of the protein that contains a high

proportion of serine, threonine, and proline (Van den Steen et al., 1998).

These residues probably enable the region of the protein to assume a

conformation that is accessible to the GalNAc transferase enzyme respon-

sible for the addition of the GalNAc. The ﬁrst step for the assembly of the

mucin type O-glycans is the addition of N-acetylgalactosamine (GalNAc)

residue to a Ser/Thr by a GalNAc transferase (GalNAcT) from UDP-

Post-translational modification of recombinant proteins 133

GalNAc (Van den Steen et al., 1998). Further elongation leads to a large

number of structures, synthesized by various glycosyltransferases, produ-

cing eight different core structures (Figure 6.4). These core structures can

be further modiﬁed by sialylation, fucosylation, sulfation, methylation, or

acetylation. The most common is the core 1 structure (Gal1-3GalNAc)

that may be monosialylated or disialylated. These are the most prominent

O-glycan structures found in glycoproteins produced from CHO cells

(Backstrom et al., 2003).

6.2.3 Patterns of glycosylation in non-mammalian cells

The prokaryotes (mainly E. coli) were the ﬁrst cells to be used for gene

expression of recombinant proteins. These cells can be manipulated

genetically and show high productivity in large-scale systems. However,

the cells lack any metabolic pathways for glycosylation and so the proteins

produced are not glycosylated. Lower eukaryotes (yeast, insect, and plant

cells) are capable of glycosylation of proteins. However, the glycans

produced in these cells differ signiﬁcantly from those present in mamma-

lian glycoproteins (Jenkins et al., 1996). Yeast, insect, plant, and mamma-

lian cells share the feature of N-linked oligosaccharide processing in the

ER, including attachment of Glc

Man

GlcNAc

-P-P-Dol and subsequent

truncation to Man

GlcNAc

structure. However, oligosaccharide proces-

sing by these different cell types diverges in the Golgi apparatus (Goochee

S/T

Core VIII

Core V

Core VI

Core VII

S/T

Core I

Core II

Core III

Core IV

Tn antigen

Linkage position

GlcNAc

Gal

GalNAc

Figure 6.4

Core structures of mucin-type O-linked glycans. These represent a series of

characteristic structures attached to a serine (S) or threonine (T) amino acid

residueofaprotein.Thereareeightcorestructures,eachofwhichmayhave

further monosaccharide additions. Tn is a precursor of the T (Thomsen–

Freidenreich) antigen which is commonly used as a marker for tumor cells.

134 Animal Cell Technology

et al., 1991). Although there is extensive heterogeneity of structures arising

from any cell type, examples of predominant N-glycans that might occur

from different systems are shown in Table 6.1.

Insect cells. Lepidopteran insect cell lines (such as Spodoptera frugiper-

da, Sf-9) have been used extensively for expression of recombinant

proteins using the baculovirus as an expression vector (Jarvis et al., 1998).

Alternative methods of protein expression are also available using efﬁcient

insect-associated promoter systems (Farrell et al., 1998). The advantage of

the use of these cells is the high expression level and growth rate of the

cells in culture. However, the glycosylation of proteins expressed by insect

cells is limited. These cells can add Glc

Man

GlcNAc

precursors to

appropriate N-glycan sites in a nascent polypeptide and convert them to

Man

GlcNAc

. They also have the enzymes necessary to trim this

oligosaccharide all the way down to Man

GlcNAc

. The formation of this

product requires the action of the enzymes ManI, GnTI, and ManII.

However, there is little structural processing beyond the Man

GlcNAc

core oligosaccharide apart from the possibility of fucosylation (Jarvis and

Finn, 1996; Donaldson et al., 1999).

Glycosyltransferase enzymes are either absent or at a low level of

activity (Jarvis et al., 1998). Therefore, generally the insect expression

system is incapable of synthesizing sialylated lactosamine complex-type

N-glycan or sialylated O-glycans. However, some insect cells have been

found to produce recombinant glycoproteins with elongated trimannosyl

core structures containing terminal GlcNAc or Gal, and one recombinant

glycoprotein acquired complex biantennary N-linked glycan containing

sialic acid (Davidson et al., 1990; Kulakosky et al., 1998).

Table 6.1 Typical predominant N-glycan structures from different c ell types.

Cell-type N-glycan Structure

Bacteria, E.coli None

-------------

Yeast High mannose

Insect Fucosylated core structure

Plant Xylosylated and fucosylated

core structure

Mammalian Complex biantennary

Post-translational modification of recombinant proteins 135

Jarvis et al. (1998) proposed a model to explain the N-linked oligosac-

charide structures found in insect cell-produced glycoproteins. In this

model the enzyme GlcNAcTII competes with N-acetylglucosaminidase

(GlcNAdase) at a branch point. Depending upon the relative activities of

these competing processing enzymes, trimannosyl core or complex

N-linked glycans may be produced. It would appear that insect cells have

a high level of N-acetylglucosaminidase activity, which can remove

GlcNAc from a terminal position. The production of complex sialylated

N-linked structures could only occur when the recombinant protein is a

very poor substrate for GlcNAdase or excellent substrate for the low level

activities of the glycosyltransferases. The enzymic activity may vary con-

siderably and this explains the differences in the N-glycan pattern of the

same glycoprotein secreted by different insect cell lines (Kulakosky et al.,

1998).

The potential of insect cells for O-glycosylation was reported in a study

in which three lepidopteran cell lines were shown to produce predomi-

nantly short O-glycan structures (Lopez et al., 1999). All three cell lines

expressed GalNAcÆ1-O-Ser/Thr (Tn antigen), whereas the ability to

synthesize Gal1-3GalNAcÆ1-O-Ser/Thr (T antigen) and GalÆ1-4Gal1-

3GalNAcÆ1-O-Ser/Thr (PT antigen) was more limited. This indicated

low activity of 1-3-galactosyltransferase and Æ1-4-galactosyltransferase.

There was no indication of sialylation of these structures, suggesting the

absence of sialyltransferase activity. The culture medium used to grow

these cell lines had a major effect on the O-glycans expressed. The use of a

semi-deﬁned rich medium enhanced glycosyltransferase activity signiﬁ-

cantly compared with a minimal nutrient medium. The limitations in the

ability of insect cells for glycosylation have restricted their use for

production of human therapeutics. Genetic engineering of these systems is

under study in an attempt to improve the glycosylation machinery to

produce more ‘‘humanized’’ glycoproteins (Jarvis et al., 1998). In particu-

lar the transformation of insect cells with the gene for mammalian 1,4-N-

acetylgalactosyltransferase with a baculovirus expression vector leads to

expression of the enzyme and results in more extensively processed N-

glycans (Jarvis and Finn, 1996).

Fungi. The early steps in the addition of carbohydrate to proteins

appear to have been remarkably conserved during evolution. The synthesis

of the Glc

Man

GlcNAc

-P-P-Dol precursor, the transfer to the polypep-

tide, and early processing in the ER are common events shared by

eukaryotic cells. However, although relatively few N-linked sites have

been characterized, it was noted that there is a trend in favor of the use of

Asn-X-Thr sites over Asn-X-Ser in yeast glycoproteins. In contrast to

mammalian cells, where several Man residues may be removed during

processing, in Saccharomyces cerevisiae, a single speciﬁc Man residue is

cleaved to form Man

GlcNAc

. Most yeast and ﬁlamentous fungi synthe-

size carbohydrate chains of the high mannose type. Complex glycan

structures are not observed among fungal glycoproteins. Addition of Man

residues to core oligosaccharides occurs very rapidly in the Golgi, forming

the characteristic high mannose structures (mannan), which can consist of

more than 50 mannose residues and resulting in high molecular weight

glycoproteins (Hersecovics and Orlean, 1993).

136 Animal Cell Technology

Proteins synthesized in yeast may also contain O-glycans consisting of

linear poly-mannose structures attached to Ser or Thr. Similar to mamma-

lian cells, O-glycosylation in yeasts has no obvious consensus sequence.

However, unlike mammalian cells O-glycosylation in yeast is initiated

with covalent attachment of mannose via a dolichol phosphate mannose

precursor. Maras et al., (1997) showed that if the high mannose structures

are trimmed in vitro by mannosidase, they can become acceptors for the

recombinant processing enzymes, N-acetylglucosaminyltransferase I,

1,4-galactosyltransferase, and Æ2,6-sialyltransferase. Mutant strains of

yeast also may synthesize truncated mannose structures.

Plants. Plant cells also conserve the early stages of N-glycosylation.

However, the processing of the oligosaccharide trimming and further

modiﬁcation of glycans in the Golgi differ from mammalian cells. Plant-

derived oligosaccharides do not possess sialic acid and frequently contain

xylose (Xyl), not normally present in mammalian N-linked oligosacchar-

ides. Typically processed N-glycans in plants have a Man

GlcNAc

structure with 1,2 xylose and/or Æ1,3 fucose residues attached to the

reducing terminal GlcNAc (Palacpac et al., 1999). Xylose is not present in

mammalian glycan structures and fucose is attached to proximal (core)

GlcNAc by an alternative linkage (Æ1,6) in mammalian cells. The presence

of these two residues (Xyl and Æ1,3 fucose) in plant recombinant glyco-

proteins and their absence in mammalian proteins makes them highly

immunogenic if present in therapeutic glycoproteins (Parekh et al., 1989;

Storring, 1992; Palacpac et al., 1999).

6.2.4 Glycosylation in animal cells: the effect of the host cell line

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 animal

species signiﬁcant variation was found in the proportion of terminal

galactose, core fucose, and bisecting GlcNAc (Raju et al., 2000).

The fact that glycoproteins normally exist as mixtures of glycoforms

suggests that the protein structure is not the primary determining factor in

glycosylation. The glycoforms that emerge from the Golgi are end

products of a series of incomplete enzymic reactions. Thus, the choice of

the host cell line is a particularly important factor in the glycoform proﬁle

of a recombinant protein (Rudd and Dwek, 1997). Sialylation patterns of

the secreted protein are particularly affected by the host cell.

6.2.5 Culture parameters that may affect glycosylation

It is important to control culture parameters to insure consistency of

glycosylation of a recombinant protein in a culture bioprocess. However,

this may not be so easy 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 continuous depletion of nutrients (particularly glucose or gluta-

mine) and accumulation of metabolic byproducts, which have been shown

Post-translational modification of recombinant proteins 137

to limit the glycosylation process (Hayter et al., 1992; Jenkins and Curling,

1994; Nyberg et al., 1999). Culture conditions such as nutrient content,

pH, temperature, oxygen, or ammonia, may have a signiﬁcant impact on

the distribution of glycan structures found on the resulting recombinant

protein (microheterogeneity). This of course is of major concern in trying

to produce consistent biopharmaceuticals. It can lead to enhanced glyco-

form heterogeneity and signiﬁcant batch to batch variation in the produc-

tion process. To maintain product quality it is important to understand the

parameters that cause the variation in glycosylation.

Glucose starvation may result in an intracellular depleted state with a

shortage of glucose-derived precursors of glycans that could result in

reduced site occupancy. This has been shown at low glucose concentration

(, 0.5 mM) for the synthesis of antibody (Stark and Heath, 1979) and

recombinant gamma-interferon (Hayter et al., 1992). Fed-batch strategies

may be designed to insure that the concentrations of these key nutrients

do not decrease below a critical level that could compromise protein

glycosylation (Xie and Wang, 1997). The lower levels of nutrients for the

production of recombinant protein from CHO cells were found to be 0.1

mM glutamine and 0.7 mM glucose (Chee et al., 2005). Nutrient levels

below these critical concentrations led to decreased sialylation and an

increase in hybrid and high mannose-type glycans. A plausible mechanism

for reduced site occupancy of a recombinant protein produced by glucose-

depleted or glutamine-depleted CHO cell cultures was offered by Nyberg

et al. (1999). They showed that in both cases the low level of glycosylation

was related to a decreased intracellular concentration of UDP-GlcNAc.

This intracellular concentration of UDP-GlcNAc has also been shown

to be important in another respect. An elevated UDP-GlcNAc appears to

cause a decrease in sialylation, which may be explained metabolically by

the inhibition of CMP-sialic acid transport (Pels Rijcken et al., 1995). This

may be the mechanism for the observed decrease in sialylation that has

been shown to accompany high levels of ammonia in culture (Yang and

Butler, 2000).

The dissolved oxygen (DO) level of cultures has also been shown to be

a critical parameter to affect glycosylation. Terminal galactosylation of an

immunoglobulin (IgG) was changed signiﬁcantly with a gradual decrease

in the digalactosylated glycans (G2) from 30% at the higher oxygen level

to around 12% under low oxygen conditions (Kunkel et al., 1998).

6.3 Other forms of post-translational modiﬁcation

6.3.1 Deamidation

Non-enzymatic deamidation of glutamine or asparagine residues of pro-

teins can occur spontaneously, leading to potential structural changes and

loss of biological activity. Deamidation of Asn occurs more rapidly and

leads to Æ-linked (Asp) or -linked aspartyl (Isoasp) via a succinamide

derivative. This introduces an extra negative charge in the protein that can

result in alterations of properties. For example, deamidation of an Asn

residue of human growth hormone-releasing factor can lead to a 25–500-

138 Animal Cell Technology

fold loss in potency (Friedman et al., 1991). Low levels of deamidation can

lead to protein aggregation and are related to amyloid formation (Nilsson

et al., 2002). The presence of Isoasp may also enhance the immunogenicity

of a protein. Deamidation may occur during bioprocessing or storage of

biopharmaceuticals and is dependent upon a number of factors including

pH, temperature, solvent dielectric constant, and ionic strength, as well as

the structure of the protein. Deamidation half-times for Asn in synthetic

peptides at neutral pH and 37

C are in the range of 1–500 days. Huang

et al. (2005) described ‘‘hot spots’’ for spontaneous deamidation in a

humanized monoclonal antibody (mAb) that can lead to loss of binding

capacity.

6.3.2 Deamination

Proteins subjected to metal-catalyzed oxidation can form carbonyl groups

that are essentially deamination products (glutamic and aminoadipic

semialdehydes) that result from proline, arginine and lysine residues. This

may lead to loss of function or structural alterations of the proteins that

are associated with the loss of a positive charge. Glutamic semialdehyde

(5 mmole/mol) accounted for the majority of carbonyl groups in proteins

from HeLa cells in culture. Treatment of the cells with an H

-generat-

ing system led to a 2.5-fold increase of these carbonyl groups (Requena

et al., 2001).

Oxidative deamination of lysine has also been observed by incubating a

model protein in the presence of high concentrations of glucose (50 mM)

or lower concentrations of the Æ-oxoaldehydes; methyl glyoxal, or

3-deoxyglucosone (1 mM; Akagawa et al., 2002). These Æ-oxoaldehydes

may be formed by the nonenzymatic degradation of glucose or from the

early stage of glycation (Thornalley et al., 1999). The presence of copper

(Cu

2þ

) increases the rate of this process.

6.3.3 Glycation

This is a non-enzymatic process of adding glycans that may contribute to

the post-translational modiﬁcation of proteins. It is a complex process that

results in lysine- and arginine-derived glycation adducts in intracellular or

extracellular proteins. The initial Maillard reaction involves the formation

of a Schiff’s base between glucose and the -amino group of lysine ( Figure

6.5). This undergoes a rearrangement to form a more stable fructosamine–

protein adduct. However, a number of subsequent reactions are possible

that lead to advanced glycation end products (AGEs) that have been

associated with pathological complications of diabetes. The process has

been well studied in connection with pathological hyperglycemia. How-

ever, there are also some speciﬁc considerations for bioprocesses.

Biopharmaceuticals may be subjected to heat treatment as a means of

viral inactivation. Protein damage is prevented by the addition of high

concentrations of a thermostabilizing excipient such as sugars or polyhyd-

ric alcohols. The non-reducing polyhydric additives increase thermostabil-

ity of the protein by the formation of a hydrogen-bonded solvent shell.

The typical treatment for factor VIII consists of heat treatment at 608C for

Post-translational modification of recombinant proteins 139

10 h in the presence of high concentrations of sucrose and glycine. Such a

treatment will destroy most viruses but may cause glycation products that

could render the protein immunogenic (Smales et al., 2002). The observed

modiﬁcations take place by reaction adducts of glucose or fructose that are

derived from the hydrolysis of sucrose. The heat treatment can result in

deamidation, protein glycation, or AGE adduct formation.

6.3.4 Gamma-carboxylation

Gamma-carboxyglutamic acid (Gla) is an amino acid with a dicarboxylic

acid side chain that has metal-binding properties and is important for the

biological activities of a range of blood coagulation proteins that include

prothrombin, factor X, factor IX, and factor VII, as well as the regulatory

proteins, protein C and protein S. The extra negative charge provided by

the carboxylation aids in the calcium-induced interaction of these proteins

with membrane surfaces. Gamma-carboxyglutamic acid is synthesized by

the post-translational modiﬁcation of glutamic acid residues (Glu) in a

reaction catalysed by a carboxylase enzyme with requirements for vitamin

R.NH glucose

⫹

CH OH

NH-protein

HOCH

Schiff’s base

CH OH

NH-protein

CH OH

Fructosamine

Amadori

re-arrangement

Figure 6.5

Glycation reactions. Glucose or other sugars may react non-enzymatically with proteins to form

glycation products. Early glycation products such as fructosamine–protein adducts may later be

transformed to the more complex advanced glycation end products (AGE products).

140 Animal Cell Technology

K, O

, and CO

(Figure 6.6).

The carboxylation of glutamic acid residues

of these proteins appears to be directed by a speciﬁc recognition sequence,

which is often found in the propeptide to insure binding to the vitamin

K-dependent carboxylase. Furie et al., (1999) identiﬁed ﬁve requirements

for gamma carboxylation:

(i) there is a recognition site that interacts with the carboxylase enzyme;

(ii) the protein is trafﬁcked through the rough ER during synthesis;

(iii) the cell has a carboxylase enzyme associated with the rough ER;

(iv) there are Glu residues within 40 residues of the enzyme recognition

sequence;

(v) intracellular vitamin K is present.

The production of recombinant protein C has been studied from BHK

cells and it was shown that the protein secretion rate increased with cDNA

copy number. However, a higher secretion rate decreased the efﬁciency of

gamma-carboxylation (Guarna et al., 1995). Similarly, culture conditions

were shown to affect carboxylation of protein C, with a good supply of

oxygen favoring high efﬁciency of gamma-carboxylation (Sugiura and

Maruyama, 1992).

Factor X undergoes extensive post-translational modiﬁcation including

gamma-carboxylation of 11 glutamic acid residues. However, the efﬁ-

ciency of this process is variable and inevitably leads to a heterogeneity of

isoforms. HEK-293 cells have been shown to be more efﬁcient for

gamma-carboxylation than CHO cells. This may depend on the availabil-

ity and activity of the carboxylase enzyme that binds to the propeptide.

Camire et al., (2000) improved the efﬁciency of carboxylation of factor X

by using a chimeric cDNA that contained a propeptide with reduced

afﬁnity to the enzyme. Surprisingly, this increased the extent of carboxyla-

tion from a normal value of 32% to 85% by enhancing substrate turnover.

Nevertheless, in all situations there were two pools of protein produced: a

fully carboxylated and uncarboxylated recombinant factor X.

OHO

Glutamic acid Gamma-carboxyglutamic acid

Carboxylase

Vitamin K,

CO , O

Figure 6.6

Gamma carboxylation of glutamic acid residue in a protein. This is an enzymic

reaction catalyzed by a carboxylase to modify a glutamic acid residue of a protein

to form gamma-carboxyglutamic acid. The carboxylase reaction requires vitamin

K, O

,andCO

Post-translational modification of recombinant proteins 141