Download 1. Introduction 2. Fundamentals 3. Glycosylation 4

Bakkalaureatsarbeit
Index
1. Introduction
2. Fundamentals
From DNA to protein
Secretory pathway
3. Glycosylation
Pichia pastoris
Glycosylation of therapeutic proteins
Differences between yeast and human N-glycosylation pathways
Humanizing glycosylation pathways in yeasts
A) Early approaches
B) Rebuilding the human glycosylation pathway
C) Galactosylation
D) Sialic acid transfer – the final step
Problems
4. Advantages and disadvantages of mammalian, yeast, fungal,
plant, and insect cells
Basis of evaluation
Mammalian cells
Yeast cells
Fungal cells
Plant cells
Insect cells
5. Summary and perspectives
6. References
Bösch Peter | 0240421
April 2005 | 1
Bakkalaureatsarbeit
Humanisation of posttranslational
modifications in yeasts
1. Introduction
Many proteins used for therapeutic reasons in medicine are not only a sequence of
amino acids determined by a specific gene, but require also trimming, editing of
amino acids, addition of sugars and many others. These alterations after the initial
translation of the proteins are called posttranslational modifications. Sometimes they
are crucial for the desired effect or enhance the effectivity of the substance ten fold or
render it useless.
Until now products for therapeutic use where expressed in mammalian cells, in
particular Chinese hamster ovary (CHO) cells. They can do all the necessary
alterations a human cell would do to produce a fully functional metabolite or enzyme.
But mammalian cells have a lot of drawbacks, so alternatives were searched for and
found in bacteria, especially Escherichia coli, yeasts like Saccharomyces cerevisiae
and Pichia pastoris, fungi, plants and insect cells. The most significant alteration
seems to be glycosylation, especially N-glycosylation.
Humanisation in this context means modification of a species; adaptation of skills not
to perform production of metabolites required in their natural habitat, but of
substances equivalent to those expressed in human cells including their
posttranslational modifications.
This text is focused on yeasts and their ability to glycosylate the proteins the human
way already or “learn” these processing steps and most important glycosylation by
the use of recombinant technology and selection so that they can be used for
production of substances for medical use.
Bösch Peter | 0240421
April 2005 | 2
Bakkalaureatsarbeit
2. Fundamentals
The description given below is based on yeast, a simple eukaryote.
Eukaryotes in contrast to prokaryotes have got compartments enclosed by a bilayer
membrane made of phospholipids.
From DNA to protein
To construct any protein, the cell requires a plan which contains all the information to
build, modify and deliver it to the desired place. This plan is called gene, which is
embedded in a double stranded string along with other genes for other proteins.
Altogether they form the genome of a cell which is in eukaryotes surrounded by a
membrane to form the nucleus. Even in a multicellular organism the genome in
different cells is always exactly the same.
These genes are made of four nucleotides organised in a double helix where the
adenine nucleotide always pairs with the thymine nucleotide and the cytosine
nucleotide always pair with the guanine nucleotide. These sequences of nucleotides
which encode a gene are copied into mRNA, where the thymine nucleotides are
replaced by uracil nucleotides, capped, polyadenylated and spliced. This process is
called transcription. Transcription is highly regulated because of its direct
responsibility for the concentration of a certain protein. Several types of cancer occur
partly because one or more regulation mechanisms are knocked out.
The reason for copying the DNA to mRNA is simply to protect the genome and when
multiple copies are made a higher level of protein is achieved in a shorter period of
time.
After the transcription the mRNA is transferred from the nucleus to the cytosol.
In the cytosol, the medium that surrounds all the inner compartments of the cell, a
machinery called ribosome binds to the mRNA. The ribosome translates the plan
encoded by the mRNA strand into a sequence of 20 different amino acids which are
the units a protein is made of. After the complete sequence is read, the protein is
released and the mRNA is degraded.
Bösch Peter | 0240421
April 2005 | 3
Bakkalaureatsarbeit
Depending on a signal sequence at the beginning of the mRNA the fate of a protein
is decided. It can either be released into the cytosol, peroxisomes, plastids,
endoplasmic reticulum (ER) or into the mitochondria. If the protein is expressed in the
cytosol it can stay there or it can be transferred into the nucleus via gated transport.
The translation/transfer of the protein into the ER is the starting point of the secretory
pathway.
Secretory pathway
The secretory pathway is mediated by vesicular transport, where membraneenclosed transport intermediates – which may be small, spherical transport vesicles
or larger, irregularly shaped organelle fragments – ferry proteins from one
compartment to another. The start point of this sequence of compartments is the ER,
followed by the Golgi, where it can either go to the late endosome and in progression
to the lysosomes where most of the substances are broken down or it buds from the
Golgi as a secretory vesicle targeted to the membrane of the cell with either
membrane proteins and other parts of the membrane or metabolites that are secreted
by the cell.
As mentioned before, a protein is expressed into the ER when it encodes a certain
signal sequence. This signal sequence is bound by a signal recognition protein which
attaches to a receptor in the ER membrane then translocation begins. When
translation and translocation are finished, the signal peptide is cleaved from the
protein. Most proteins synthesized in the rough ER are glycosylated by the addition of
an N-linked oligosaccharide. The sequence of the amino acids and the addition of the
oligosaccharides force the protein to fold into its native form, oligomerize and form
disulfide bonds. If not folded or oligomerized correctly, they are exported to the
cytosol, where they are deglycosylated, ubiquitylated, and degraded in proteasomes
[1].
The correctly folded and assembled proteins bud from the ER and reach the cisGolgi.
The Golgi finishes the glycosylation that has started in the ER by cleavage and
addition of sugars, phosphorylaton of oligosaccharides on lysosomal proteins and
Bösch Peter | 0240421
April 2005 | 4
Bakkalaureatsarbeit
sulfation of tyrosines and carbohydrates. The proteins that leave the trans-Golgi are
sorted and transported to the lysosome, the plasma membrane or the secretory
vesicle.
The next chapter introduces the most important yeasts used for glycosylation
engineering so far and covers the glycosylation of proteins in more detail, the
necessary processing steps in yeast and in humans, their importance in
pharmaceutical products and the problems that occur by transforming the yeast
based glycosylation pathway into a humanlike pathway.
Bösch Peter | 0240421
April 2005 | 5
Bakkalaureatsarbeit
3. Glycosylation
Pichia pastoris
The methylotrophic yeast Pichia pastoris is now one of the standard tools used in
molecular biology for the generation of recombinant protein. P. pastoris has
demonstrated its most powerful success as a large scale (fermentation) recombinant
protein production tool. What began more than 20 years ago as a program to convert
abundand methanol to a protein source for animal feed in the form of single cell
proteins has been developed into what are today two important biological tools: a
model eukaryote used in cell biology research and a recombinant protein production
system. In the controlled environment of a fermentor it is possible to grow the
organism to high cell densities (>100g/l dry cell weight), where the product in the
medium, the secreted proteins is roughly proportional to cell mass. Significant
advances in the development of new strains and vectors, improved techniques, and
the commercial availability of these tools coupled with a better understanding of the
biology of Pichia species have led to this microbe’s value and power in commercial
and research labs alike [2].
The Pichia pastoris system for expression of heterologous recombinant proteins is
being used increasingly because of the large yields of properly folded – processing of
signal sequences (both pre- and prepro-type), disulfide bridge formation - proteins
that result and the ease of scaling up pilot protocols into large-biomass fermentors.
Another advantage of this system centres on the type of glycosylation that result,
generally yielding protein-bound oligosaccharides of O- and N-linked type that are of
much shorter chain length than found in Saccharomyces cerevisiae [3].
With Pichia pastoris as expression system, yields of up to 15g/l (gelatin production)
were achieved, where production outcomes of >1g/l are termed as high protein titers.
For that reasons the text is focused on Pichia pastoris where major breakthroughs
have been reported in the last years and which is the best characterized system to do
humanization of yeast by the time of writing.
Bösch Peter | 0240421
April 2005 | 6
Bakkalaureatsarbeit
Glycosylation of therapeutic proteins
As mentioned earlier, several hundred proteins have been expressed recombinanly
in the last couple of years in yeast. By now, the host system wasn’t able to do some
of the most critical modifications so a lot of highly interesting proteins such as some
antibodies, EPO, Serine protease inhibitor [4], hepatitis B [5] vaccine, Antithrombin III
[6] and many more couldn’t be fully considered for production in yeast systems so far.
This is mostly, because yeast does N-glycosylation of the high-mannose type or also
called hypermannosylation with up to 40 mannose residues. Human cells do not
hypermannosylate but use several additional types of sugars to synthesize an Nglycan of the complex type. So if a human glycoprotein is expressed in yeast, the
sequence of the amino acids will be right, even the location of the glycosylation will
be right, but the glycan isn’t of the complex type but of the high-mannose type. The
best case scenario will lead to a fully functional glycoprotein which is not recognized
by the immune system of the patient it is applied on to. Most of the time the highmannose part is recognized by mannose receptors present on macrophages and
endothelial cells. The substance will show an extremely short half life and may not be
able to do any effect in its target tissue. If the glycosylation is skipped altogether the
protein will misfold (structural importance of glycosylation) and so is going to be
biologically inactive as well as cleared from circulation. Therefore no glycosylation at
all or the wrong type of glycosylation is not an option.
By now, the way to go was mammalian cells, especially Chinese hamster ovary
(CHO) cells as expression system. Although the glycosylation doesn’t match a
hundred percent, only minor differences have been found which can be eliminated by
modification to achieve maximum efficacy [7].
But there are certain drawbacks with mammalian cell cultures. They grow slowly and
are very demanding. Even slight changes in system parameters can slash the
production or lead to other metabolites which require an additional step in down
stream processing after the fermentation. Mammalian cell cultures do not secrete
homologous glycoproteins but a heterogeneous, mixture of several glycoforms. Every
single glycoform has got its own kinetics and for approval after clinical phases these
proportions have to stay the same. The scale up from small cell cultures to big
fermentors is highly critical if sometimes simply not possible.
Bösch Peter | 0240421
April 2005 | 7
Bakkalaureatsarbeit
To overcome these drawbacks one may attempt to further improve the mammalian
system, which is already on its limit or take another way.
This means, to change the production system radically and search for other options.
This will be discussed in Chapter 4, where the advantages and problems of other
systems are discussed in detail.
In the meantime the focus is on yeast as production system, the difference to
thehuman glycosylation pathway and the way it can be adapted.
Differences between yeast and human N-glycosylation pathways
The starting point either for human or yeast glycosylation is always the endoplasmic
reticulum. There, a preassembled core oligosaccharide (Glc3Man9GlcNAc2) bound
to the membrane component dolichol is transferred onto the nascent polypeptide
which has been imported into the ER because of its signal sequence (see Chapter 2:
the secretory pathway).
This step is catalyzed by an enzyme complex named
oligosaccharyltransferase and has been conserved through all evolutionary steps
between yeast cells and a human cell. The transfer depends on the recognition of an
amino acid triplet which sequence is Asn-X-Ser/Thr (where X is any amino acid other
than proline). The core oligosaccharide is attached to the asparagine residue in this
sequence.
After these initial steps, the core oligosaccharide has to undergo several enzymatic
modification steps. First two enzymes called glucosidase I and glucosidase II cut
away the glucose sugars at the end of the branches which initiates a process called
glycan-mediated chaperoning. The last sugar that is removed in the ER is a
mannose, trimmed by a α-1,2-mannosidase. The glycoprotein leaving the ER has
always the same structure (Man8GlcNAc2). Incorrectly folded proteins are
reglucosylated transported into the cytosol for degradation. Therefore this process is
important for quality control, otherwise misfolded proteins might be secreted.
The resulting Man8GlcNAc2-containing glycoprotein is then transported to the Golgi
apparatus where the N-glycan takes different routes in yeast and humans. In
mammals the next step is cleavage of 3 mannose sugars by a α-1,2-mannosidase
which results in Man5GlcNAc2, the substrate for N-acetylglucosaminyltransferase I,
Bösch Peter | 0240421
April 2005 | 8
Bakkalaureatsarbeit
which adds a single N-acetylglucosamine (GlcNAc) sugar onto the terminal of the
1,3-mannose.
Figure1 | Major N-glycosylation pathways in humans and yeast. Representative pathway
in humans (left) provides a template for humanizing N-glycosylation pathways in yeast (right).
ER, endoplasmic reticulum; GalT, galactosyltransferase; GlcNAc, N-acetylglucsoamine;
GnT I&II, N-acetylglucosaminyltransferase I & II; Man, mannose; Mns II, mannosidase II;
MnTs, mannosyltransferase; NANA, N-acetylneuraminic acid; ST, sialyltransferase. [8]
Mannosidase II removes the last two mannose sugar terminals which results in
GlcNAcMan3GlcNAc2. N-acetylglucosaminyl transferase II adds one GlcNAc sugar
and galactosyltransferase adds two galactose sugars. The final processing step is
catalysed by sialyltransferase where an N-acetylneuraminic acid (NANA) is
transferred to every terminal of the two branches. The product of the human
glycosylation pathway after ER and Golgi is NANA2Gal2GlcNAc2Man3GlcNAc2 also
known as complex type. In addition, there are other glycosyltransferases such as
GalNAc transferases, GlcNAc phosphotransferases and fucosyltransferases which
allow the spectrum of glycosylation to be much wider.
Bösch Peter | 0240421
April 2005 | 9
Bakkalaureatsarbeit
Yeast alters the structure of the basic glycan exported from the ER with
mannosyltransferases which add mannose sugars as well as mannosylphosphate
transferases which leads to a hypermannosylation of the glycoprotein with varying
extents depending on the organism which gives a range of high-mannose type
oligosaccharides of Man9…100GlcNAc2.
In the end, the oligosaccharides are secreted either constitutively or regulated.
Constitutive secretion is done all the time and in balance with endocytosis. Regulated
secretion happens when a signal molecule triggers the release of the metabolite, for
example the release of the neurotransmitter acetylcholine.
Pichia pastoris’ glycan structures are typically smaller then those of the yeast
Saccharomyces cerevisiae and it has got a secretory pathway with distinct Golgi
stacks similar to those found in mammals [8].
Humanizing glycosylation pathways in yeasts
In the previous section the differences of the glycosylation pathways between yeast
and human cells were described. This section will cover all the steps required to
create a yeast cell enabled of doing complex N-glycosylation.
A) Early approaches
After the glycosylation pathways of both species were understood, or at least partially
unravelled strategies for achieving the transformation of a yeast where devised. The
most promising was knocking out the complete pathway in yeast that doesn’t match
the human one and substitute it with human enzymes. Because of the popularity and
well understood pathways in Saccharomyces cerevisiae early approaches targeted
that organism, although it has a much higher degree of mannosylation then Pichia
pastoris. The idea was to eliminate the α-1,6-mannosyltransferase Och1p which is
the core point of the outer chain elongation [8]. Jigami et al were the first who
attempted
a
knockout
of
both,
the
Och1
gene
which
encodes
α-1,6-
mannosyltransferase and Mnn1 which encodes α-1,3-mannosyltransferase resulting
in mostly Man8GlcNAc2 type oligosaccharides [9]. Man8GlcNAc2 is the starting point
of the human glycosylation pathway. Chiba et al figured out in 1998 how to remove
three mannose sugars after having the Man8GlcNAc2 substrate transported into the
Bösch Peter | 0240421
April 2005 | 10
Bakkalaureatsarbeit
Golgi. They took a α-1,2-mannosidase from Aspergillus saitoi and placed it in the ER
of Saccharomyces cerevisiae. This led to a yield of 27% Man5GlcNAc2 the rest was
higher mannosylated [10]. By this time most of the attempts to humanize a yeast cell
failed because either a strain was chosen which doesn’t provide an appropriate
substrate or the elimination of the original N-glycosylation pathway in a way that it
doesn’t compete with the newly engineered on was difficult.
B) Rebuilding the human secretory pathway
Building a glycan is a step by step process that requires processing in a distinct
order. For example mannosidase II can only trim terminal α-1,6-mannose and α-1,3mannose if a terminal GlcNAc is present on the α-1,3 arm. The enzymes for Nglycosylation have therefore to be arranged in a specific order along the ER / Golgi
pathway to be able to glycosylate efficiently and correctly.
So after overcoming the first obstacle by eliminating the old pathway and providing
the substrate for human glycosylation which is Man8GlcNAc2 the next milestone was
to arrange all the different mannosidases and glycosyltransferases along the way to a
glycan of the complex type in the right order on the right place.
The sequence of compartments a glycoprotein has to undergo is ER to cis-, medial –
and trans-Golgi compartment from where the product is transported to the membrane
where it is secreted. Most
glycosyltransferases and mannosidases
such as α-1,2-mannosidase,
mannosidase II and GlcNAcT I are
anchored to the membrane through
a type II transmembrane anchor with an
N-terminal transmembrane domain.
The short N-terminus in the cytosol
and the nearby transmembrane region
encode the targeting information.
Figure2 | Type II membrane proteins. The anchoring mechanism for
most glycosyltransferases is through a type II transmembrane anchor,
by which the C-terminal catalytic domain is anchored to the membrane
through an N-terminal transmembrane domain. [8]
Bösch Peter | 0240421
April 2005 | 11
Bakkalaureatsarbeit
This finding was taken advantage of by the laboratory of Wildt and Gerngross [11].
Because it is not reliably predictable which sequence encodes targeting information
for a certain Golgi subcompartment and because the same sequence will lead to
different location of the enzyme in diverse species two libraries were built, one with
N-terminal fragments of ER and Golgi enzymes of Saccharomyces cerevisiae and
one with catalytic domains of several α-1,2-Mannosidases from Homo sapiens, Mus
musculus, Aspergillus nidulans, Caenorhabditis elegans, Drosophila melanogaster
and Penicillium citrinium, but without N-terminus.
Those two libraries were combined, every yeast N-terminus with each catalytical
domain of different species. That resulted in 608 so called chimeric fusion proteins.
As host was chosen a Pichia pastoris strain that lacks α-1,6-mannosyltransferase
activity but is able to secrete a hexahistidine-tagged fragment of human plasminogen
as a reporter protein. Only few strains were able to trim Man8GlcNAc2 to
Man5GlcNAc2.
Next step was the addition of an N-acetylglucosamine by N-acetylglucosaminyl
transferase I. The obstacles to overcome this problem were the requirement for
Man5GlcNAc2 which is the substrate, then the right localization of the enzyme and
the availability of UDP-GlcNAc in situ. Again a set of libraries was made with addition
of a UDP-GlcNAc transporter from K. lactis for efficient UDP-GlcNAc transport. That
experiment yielded strains that where able to produce GlcNAcMan5GlcNAc2.
The GlcNAcMan5GlcNAc2 complex is substrate for the next enzyme, the
mannosidase II, which removes one terminal α-1,3- and one terminal α1,6-mannose.
That yields GlcNAcMan3GlcNAc2. The N-acetylglucosamine added in the step
before is essential; otherwise the reaction wouldn’t take place. This underlines the
importance of an approach that is similar to the order of the human glycosylation
pathway.
Also N-acetylglucosaminyl transferase II was introduced into P. pastoris.
This
enzyme extends the branch without N-acetylglucosamine so that the three mannose
sugar core is extended by an N-acetylglucosamine on each branch that is
GlcNAc2Man3GlcNAc2 [8].
Bösch Peter | 0240421
April 2005 | 12
Bakkalaureatsarbeit
Figure3 | A working model for the cellular distribution of glycosyltransferases
throughout the secretion pathway. Specific glycosyltransferases and glycosidases line the
luminal surface of the endoplasmic reticulum (ER) and Golgi, allowing the sequential processing
of glycoproteins as they are shuttled through the secretory pathway. GalT, galactosyltransferase;
GnT, N-acetylglucosaminyltransferase; ST, sialyltransferase; TGN, trans-Golgi network. [8]
C) Galactosylation
The previous step provided a paucimannose core (Man3GlcNAc2) plus an Nacetylglucosamine on every branch. A galactosyl transferase should add one
galactose sugar onto every branch.
This part was rather tricky because of problems other groups also encountered
working with Saccharomyces cerevisiae [12].
The group around Wildt and Gerngross reported two approaches where only one is
described here. By expressing only galactosyl transferase the UDP-galactose would
be missing. So in addition, a UDP-galactose-4-epimerase was needed, that would
convert UDP-glucose from the cytosol into UDP-galactose. This enzyme was taken
out of S. pombe. Yet the UDP-galactose was in the cytosol, and the glycan and the
Bösch Peter | 0240421
April 2005 | 13
Bakkalaureatsarbeit
galactosyl transferase are in the Golgi. From D. melanogaster a UDP-galactose
transporter was co-expressed in the Pichia pastoris strain.
In the Golgi all 4 components meet. The substrate (the GlcNAc2Man3GlcNAc2
synthesized earlier), the UDP-galactose made by UDP-galactose-4-epimerase out of
UDP-glucose present in the cytosol and transported from there to the Golgi by the
UDP-galactose transporter and the galactosyl transferase form a complex which
product is a Gal2GlcNAc2Man3GlcNAc2 biantennary glycan.
D) Sialic acid transfer – the final step
The last and probably most critical step is the transfer of sialic acid to each of the
antennarey of our glycoprotein. The method is the same as with earlier enzymes.
First a library with catalytic domains and a library with several N-terminal fragments
for the location are combined and screened. Several attempts for introducing
sialyltransferase have been successfull.
Second a pool of CMP-sialic acid has to be available. That is a major problem,
because there is no sialic acid available in yeast. Therefore a pathway to produce
CMP-sialic acid and a CMP-sialic acid transporter have to be introduced to the
system. Already there are some preliminary data available though not published yet
[8]. Therefore it is difficult to judge the over all outcome and yield.
Problems
Although the first major breakthrough has been made by Wildt et al by reconstructing
the N-glycosylation pathway of humans in the yeast Pichia pastoris, there is a long
way to go.
A problem is that also additional glycosyltransferases such as GalNAc transferases,
GlcNAc phosphotransferases and fucosyltransferases modify the final product.
So each glycoprotein has to be expressed within a specialized strain that can do the
specific alteration.
Further is not known, how effective this system works. James Cregg talks of a 50 to
75% probability of expressing any protein of interest in Pichia pastoris at a
reasonable level after having succeeded in generating an initial breakthrough such
as Gerngross et al [7]. This is concluded because the system then can be optimized
Bösch Peter | 0240421
April 2005 | 14
Bakkalaureatsarbeit
by a couple of well known parameters which promise to yield more product. As
mentioned earlier up to 15g/l have been achieved so far, but a yield of >1g/l is
already excellent for an eukaryotic expression system [2].
Yeasts are well known to be robust and easy to scale up. But after having made so
many changes in the secretory pathway through recombinant protein technology it
might have lost a lot of it previous toughness. That is because the membrane is also
modified in this pathway and many of its proteins are glycosylated. But that has to be
proven yet.
Bösch Peter | 0240421
April 2005 | 15
Bakkalaureatsarbeit
4. Advantages and disadvantages of mammalian,
yeast, fungal, plant, and insect cells
This text is primarily focused on yeast and its capacity to replace mammalian cells for
human protein expression. But there are other competitors in this sector with their
own advantages and drawbacks.
In this chapter these are summarised for a better overall view and to understand how
the current market for such systems has evolved or will. Escherichia coli and other
bacteria are not covered because of their total lack of compartments, folding
problems and the lack of higher glycosylation systems.
Basis of evaluation
If a novel protein has to be evaluated, there are five parameters [8] that have to be
considered before attempting:
•
the cost of manufacturing and purification
A plant that can be grown anywhere and eaten without any purification step is much
cheaper than mammalian cells, which have to be passaged every other week and
require a high amount of purification steps to be below the tolerance for viral
contamination and other criteria of purity.
•
the ability to control the final product including its posttranslational
processing.
Yeast are meant to do homogeneous glycosylation, mammalian cells don’t, yet
another important factor to slash costs on down stream processing.
•
the time required from gene to purified protein.
That’s a big one with mammalian cells, because of the extremely long time a lot of
money is required and a rerun of the complete pipeline takes months.
Bösch Peter | 0240421
April 2005 | 16
Bakkalaureatsarbeit
•
the regulatory path to approve a drug produced on a given expression
platform
The approval of a drug is a time consuming and money robbing undertaking with
many hurdles. It accounts for the biggest junk of costs when developing a new
therapeutic agent. Proprietary expression systems are likely to encounter much
higher approval standards than for example a mammalian system that has been here
for decades.
•
the overall royalties associated with the production of a recombinant
product in a given host.
Most of the time one or more patents have already been filed by other companies
while developing a new system and so can cause costs before the project has even
kicked of. This has to be considered because they can easily weigh out the merits.
These 5 points are considered for the evaluation of different expression systems [8].
Mammalian cells
Most of the critical points have been mentioned in the description of the key points.
Mammalian cells are expensive to maintain because they need a complex medium.
This is also a problem, because the growth factors required are added by calf serum
which can include contamination with viruses or prions. When the protein is secreted,
the downstream process is cheaper then with inclusion bodies that have to be
refolded. The product that is secreted is not homogeneous like that of yeast but
contains heterogeneous glycans.
The production cycle of a mammalian cell is extremely long and the yield is very
small compared to yeast.
Because a lot of drugs produced by mammalian cells have been approved, it’s rather
easy to push through a new substance based on the same systems, which has been
validated before. The idea of production with mammalian cells is not new and it’s
likely that a lot of patents have been developed on this field which directly leads to
royalty fees when considered.
Bösch Peter | 0240421
April 2005 | 17
Bakkalaureatsarbeit
Yeast cells
Yields of up to 15g/l can be achieved with this system, which secrets its products.
Therefore the substance is found in the medium surrounding the cells which needn’t
to be destroyed for purification. In our case the glycoprotein that is secreted is
homogeneous. For that reason several critical and very expensive purification steps
can be avoided.
The time that an expression system with the right posttranslational modifications is
available to the time where a product is finally expressed is very short because of the
fast growth of yeast cells.
To obtain approval for glycoprotein made by yeast is another issue, because no
glycoprotein expressed by yeast has been yet approved; stringent qualification
criteria should be considered possible.
Glycoproteins made by yeast is a new idea, little work in this huge field has been
done by now; royalties are not a major concern right now.
Fungal cells
Fungal cells have pretty much the same properties as yeast cells, when it comes to
production costs. It should be mentioned that fungal cells have got much higher rates
of secretion then a yeast which is a big advantage because the concentration in the
medium is much higher which results in a cheaper down stream process.
Plant cells
Vaccines made by plants which can be taken in orally have the potential of solving a
lot of health issues in countries of the third world. They can be grown normal fields
like any other crop, stored for many months without the need for a fridge that keeps
the classical vaccines intact and taken in orally without the need of a doctor and
sterile conditions. Additionally plants can be cultivated in fermentors, but they grow
much slower then yeasts or fungi. The overall time from gene to product is a big
drawback because the cultivation takes place over many months.
A lot of transgenic plants have been released to the marked so there is enough
knowledge about getting a new hybrid seed approved. Many companies have
invested big money in researching transgenic plants.
Bösch Peter | 0240421
April 2005 | 18
Bakkalaureatsarbeit
It’s unlikely that new products come without royalties.
Insect cells
The insect cell system is the closest system to the mammalian cells.
Recent reports have shown that insect cells can be grown without serum, so viruses
and prions are no longer an issue. But the medium is still very complex which makes
it more expensive then medium for yeasts.
Insect cells can secrete the desired product so down stream processing is
approximately the same as yeasts.
If the cells are grown in serum free medium, approval gets much easier and the down
stream process is much cheaper because of no additional steps to yield a higher
level of pureness.
Bösch Peter | 0240421
April 2005 | 19
Bakkalaureatsarbeit
5. Summary and perspectives
For several years now, proteins for therapeutic reasons are expressed in yeast. That
is for example Insulin by NovoNordisk, Hepatitis B surface antigen by
GlaxoSmithKline, Merck and several others, Glucagon by NovoNordisk and many
more. They sum up to about 140 approved therapeutic proteins, with another 500 in
clinical trials [13]. Mammalian cells, the classic production system for recombinant
human glycoproteins, are able to do all the necessary posttranslational modifications
especially N-glycosylation of the complex type. N-glycosylation is of importance
because of its influence on half-life and hydrodynamic volume of the protein. If the
metabolite is glycosylated the wrong way or not at all, it may never develop a
complete level of impact.
But mammalian cells have a lot of drawbacks.
First they grow slowly, very slowly. Long generation times make minor alterations a
big issue. Their continuous need for passaging them take up a lot of manpower and
is a risk because of contamination.
Second the medium is very complex. They just won’t grow in any medium. They need
certain growth factors derived from animals such as calf serum, which can be a
problem because of the risk of introducing viruses or prions.
Third the scale up of mammalian cells can be a big issue because of the
heterogeneous mixtures of glycoforms that can be a consequence.
All these factors make this system very unattractive and expensive for expression of
recombinant proteins. But by now, it was the only way to go.
Because of this bottleneck, a lot of research on this particular field has been done to
find alternatives. One of the most promising is yeast. It is robust, the medium is well
defined and cheap, scale up is easy and genetic engineering has been done on
yeast for decades.
In the last couple of years groundbreaking discoveries have been made in the sector
of reconstructing the human N-glycosylation pathway in Pichia pastoris, an organism
used for the expression of several non-glycosylated recombinant therapeutic proteins
with outstanding features such as yields of up to 15g/l.
Bösch Peter | 0240421
April 2005 | 20
Bakkalaureatsarbeit
This year, very promising results have been reported about having achieved a
complete reconstruction of the human N-glycosylation pathway in that yeast by Wildt
et al [8].
Experiences with yeast show, that once the major breakthrough has been made, the
optimisation and scale up of the system is quite easy.
This can directly lead to glycoproteins for therapeutic reasons in nearer future such
as human erythropoietin which is used for patients with anaemia or by athletes to
increase their performance. Other products are antibodies, such as Herceptin and
Rituxan, interferon-β for the treatment of multiple sclerosis and glucocerebrosidase.
While the production cycle from the start over fermentation and downstream
processing to the final product in yeast is much cheaper and shorter than in
mammalian cells, the clinical trial for approval of a drug is still the most expensive
step which has to be taken by either of the two systems. Therefore no big price drop
in therapeutic agents is anticipated because of conversion to yeast secreted proteins.
But unlike mammalian cells, yeasts secrete homogeneous glycoproteins which
characteristics can be researched in more detail than the mixture of glycoforms
obtained in mammalian expression systems.
Therefore, yeast based expression systems for glycoproteins have a very good
chance to take over a big junk of the products made by mammalian cells right now.
Bösch Peter | 0240421
April 2005 | 21
Bakkalaureatsarbeit
6. References
1
Bruce A., Johnson A., Lewis J., Raff M., Roberts K., Walter P., (eds.),
In Molecular Biology of the Cell (4th ed.), Garland Science (2002),pp.665-709
The Cell, 665-709
2
Cregg J. M., Cereghino J. L., Shi J., Higgins D. R., ( 2000 )
Recombinant Protein Expression in Pichia pastoris
Molecular Biotechnology 16, 23-39
3
Bretthauer R. K., Castellino F. J., (1999),
Glycosylation of Pichia pastoris-derived proteins
Biotechnology and Applied Biochemistry 30, 193-200
4
Macauley-Patrick S., Fazenda M. L., McNeil B., Harvey L. M., (2005)
Heterologous protein production using the Pichia pastoris expression system
Yeast 2005, 22, 249-270
5
Martinet W., Maras M., Saelens X., Jou W. M., Contreras R., (1998)
Modification of the protein glycosylation pathway in the methylotrophic yeast
Pichia pastoris
Biotechnology Letters, 20, 1171-1177
6
Mochizuku S., Hamato N., Hirose M., Miyano K., Ohatani W., Kameyama S.,
Kuwae S., Tokuyama T., Ohi H., (2001)
Expression and Characterization of Recombinant Human Antithrombin III in
Pichia pastoris
Protein Expression and Purification, 23, 55-65
7
Tillman U Gerngross, (2004)
Advances in the production of human therapeutic proteins in yeasts and
filamentous fungi
Nature Biotechnology, 22, 1409-1414
8
Wildt Stefan, Tillman U. Gerngross (2005)
The humanization of N-glycosylation Pathways in yeast
Nature Reviews in Microbiology, 3, 119-128
Bösch Peter | 0240421
April 2005 | 22
Bakkalaureatsarbeit
9
Nakanishi-Shindo, Y., Nakayama, K., Tanaka, A., Toda, Y., Jigami Y., (1993)
Structure of the N-linked oligosaccharides that show the complete loss of α1,6-polymannose outer chain from Och1, Och1 Mnn1, and Och1 Mnn1 Alg3
mutants of Saccharomyces cerevisiae
Journal of Biological Chemistry 268, 26335-26345
10
Chiba, Y. et al. (1998)
Production of human compatible high mannose-type (Man8GlcNAc2) sugar
chains in Saccharomyces cerevisiae
Journal of Biological Chemistry 273, 26296-26304
11
Choi B., Bobrowicz P., Davidson R. C., Hamilton S. R., Kung D. H., Li H.,
Miele R. G., Nett J. H., Wildt S., Gerngross T. U. (2003)
Use of combinatorial genetic libraries to humanize N-linked glycosylation in the
yeast Pichia pastoris
Proceeding of the National Academy of SciencePNAS USA, 100, 5022-5027
12
Roy S. K., Yoko-o T., Ikenage H., Jigami Y. (1998)
Functional evidence for UDP-galactose transporter in Saccharomyces
cerevisiae through the in vivo galactosylation and in vitro transport assay
Journal of Biological Chemistry 273, 2583-2590
13
Walsh G., (2004)
Biopharmaceutical Benchmarks – 2003
Nature Biotechnology, 21, 865-870
Bösch Peter | 0240421
April 2005 | 23