Modifying glycosylation

1

• Modifying glycosylation
• Add or subtract sites to your favorite protein
  (cis)
• 1a. Subtract sites Easy, change N or S or T to A
  by site-directed mutagenesis
• 1b. Add sites Not so easy.
• Consensus N-X-S does not work, e.g.
• requires the insertion of a 12 aa region
  encompassing
• a real N-glycosylation site (6 suffices for
  O-linked)
• Place on an end or on a loop (must know
  proteins structure)
• Works
• Change the general glycosylation phenotype of the
  host cell (trans)
• E.g., Pam Stanley lectin-resistant mutants

Recorder on?
2

• Modifying glycosylation
• Add or subtract sites to your favorite protein
  (cis)
• Change the general glycosylation phenotype of the
  host cell (trans)

Clone enzyme genesGlycosyl transferases,
mostlyAlso some synthetases (e.g., NAcNeu, i.e.,
biosynthetic path to sialic acid) Can be
complexe.g., 7 different fucosyl transferases
(FTs),with different (overlapping) substrate
specificities Simpler example Hamster cells
do only 2,3 sialylation. Humans do 2,6 as well,
via a 2,6 sialyl transferase (ST) ExperimentOve
r-express cloned human 2,6 ST, along with a
substrate protein.producing permanent
transfectants of CHO cells Works Get both types
of structures now, substantially (although not
exactly the same ratio as in human cells).
Zhang, X., Lok, S.H., and Kon, O.L. 1998. Stable
expression of human alpha-2,6-sialyltransferase
in Chinese hamster ovary cells functional
consequences for human erythropoietin expression
and bioactivity. Biochim Biophys Acta 1425
441-452.
3
Isolate mutant mammalian cell lines deficient in
specific glycosylation enzymes
Stanley Isolation of multiply mutated
glycosylation mutants by selecting for lectin
resistance Lectins carbohydrate-binding
proteins Plant lectins used mostly here (but
occur widely in animals as well) Sequential
selections, push - pull on resistance,
sensitivity Resistance enzyme deficiency ?
failure to add the sugar need for lectin
binding Increased sensitivity failure to add a
sugar produces greater exposure of underlying
sugars in a transferase-negative mutant ?
better binding to the exposed sugar Showed power
of selection Showed usefulness of
complementation analysis via cell
hybridization Hybrid selection All lec-R
mutants were WGA (wheat germ agglutinin)
resistant (various degrees) pro- Tester
parent was single lec-R GAT- (req. glycine,
adenine and thymidine) Select in medium lacking
pro, GAT, and with /- WGA Complementing
hybrids will have regained sensitivity to
WGA Mutants in the same gene will remain WGA
resistant (non-complementation) Could now be
used as a tabla rasa (blank slate) for the
introduction of a series of enzymes to build
custom-tailored glycoconjugates. Complicated
though (order of addition, location in the Golgi,
etc. ) Potential targeting to
carbohydrate-sensitive receptors (e.g., liver
asialoglycoprotein receptor) clearance rate
Pam Stanley
Review Nature Biotechnology  19, 913 - 917
(2001) , The bittersweet promise of glycobiology.
Alan Dove
4
Umana, P., Jean-Mairet, J., Moudry, R., Amstutz,
H., and Bailey, J.E. 1999. Engineered glycoforms
of an antineuroblastoma IgG1 with optimized
antibody-dependent cellular cytotoxic activity.
Nat Biotechnol 17 176-180.
Target here (bisecting NAcGlc)
(NAcG N-acetyl-glucosamine here)
Presence of the bisecting NAcGlc enhances binding
of T-cell receptor to the Fc region of
antibodies. Binding is needed for ADCC. Mouse
and hamster cell lines used for commercial
production lack the glycosyltransferase needed
for bisecting NAcG addition A rat myeloma cell
line does produce MAb with the bisecting
NAcG. Hypothesis Expression of the rat enzyme
in a CHO cell line will add a bisecting NacG to
the anti-neuroblastoma MAb produced by these
cells. The modified MAb will be a better
mediator of ADCC. Experiment Clone the cDNA
for this enzyme from the rat line and transfer it
to CHO cells, driven by an inducible tet
promoter. Check sugar structure of MAb and ADCC
efficiency of the MAb. Both improved.
5
TARGET CELL
(Killer T-cell)
Genentech
Commercial MAb injected as a therapeutic
T-cell surface receptor binds Fc region of
antibody molecule(Fc gammaR)
6
Protein Glycosylation Knock-out
Assigned
Naoko Yamane-Ohnuki, et al..  Establishment of
FUT8 knockout Chinese hamster ovary cells an
ideal host cell line for producing completely
defucosylated antibodies with enhanced
antibody-dependent cellular cytotoxicity.  
Biotechnol Bioeng. 2004 Sep 587(5)614-22
Optional Update Kanda Y, Yamane-Ohnuki N, Sakai
N, Yamano K, Nakano R, Inoue M, Misaka H, Iida S,
Wakitani M, Konno Y, Yano K, Shitara K, Hosoi S,
Satoh M.  Comparison of cell lines for stable
production of fucose-negative antibodies with
enhanced ADCC.  Biotechnol Bioeng. 2006 Jul
594(4)680-8.
Other refs.
Review Grabenhorst, E., Schlenke, P., Pohl,.,
Nimtz, M., and Conradt, H.S. 1999. Genetic
engineering of recombinant glycoproteins and the
glycosylation pathway in mammalian host cells.
Glycoconj J 16 81-97.
Background Stanley, P. 1989. Chinese hamster
ovary cell mutants with multiple glycosylation
defects for production of glycoproteins with
minimal carbohydrate heterogeneity. Mol Cell
Biol 9 377-383.
7
Hypothesis Fucose interferes with binding of
the T-cell Fc-gamma-3 receptor to the Fc region
of an antibody molecule. Elimination of fucose
from produced MAbs will increase ADCC Create a
mutant CHO cells (starting with amplifiable dhfr-
cells) in which the fucose trasnferase genes have
been knocked out. All MAbs produced in these
mutant cells will be better at promoting ADCC
8
Double knock-out strategy for FUT8 an
alpha-1,6,fucosyl transferase
Little sequence data available for Chinese
hamster Isolate CHO cDNA using mouse sequence
data for primers Use CHO cDNA to isolate CHO
genomic fragments from a commercial lambda library
K.O. exon 1 translation start region
Homology regions
DT diphtheria toxin gene, Kills if integrated
via non-homologous recombination
For hemizygote Select for G418
resistance, Screen by PCR for homologous recomb.
108 cells ? 45000 colonies? 40 false
recombinants (extension-duplications) 1 true
recombinant
Step 2 for homozygote, select for
Pur-resistance 1.6X108?70,000 screened ? 10
double KO homozygotes.
Lox sites for later removal of drug resistance
marker
Remove drug resis. genes by transient
transfection with Cre Recombinase. Exon 1 has
undergone a 200 nt deletion.
Note 10s of thousands of PCRs performed to
screen for homologous recomb., using 96-well
plates
9
Double knockout evidence
After Cre treatment
Original KOd genes have a 1.5 kb
insertion (Southern blot)
Final mRNA has 200 nt deletion (RT-PCR
10
Use of a fluoresceinated lentil lectin (LCA) that
binds fucose oligosaccharides to demonstrate lack
of fucosylation in glycosylated proteins in the
FUT8 -/- cells
Control background fluorescence(FL-anti avidin)
FUT8 /
FUT8 /-
Surprising that fucosylation levelis down by
one-half that CHO cells do not have excess
fucosylation capacity . . .
FUT8 -/-
11
Rituxan (anti-CD20) produced in FUT -/- cells
does not contain fucose
(CD20 is expressed at a high level on the surface
of many lymphoma cells.)
HPLC analysis
Digestion all the way to monosaccharides
Missing d - g
12
Binding to CD20 membranes is the same FUT8-/-
anti CD20 Ritxuan
In ADCC, FUT8-/- anti-CD20 Rituxan
Anti-CD20 from a partially FUT-deficient rat cell
line
Characteristics ofFUT8-/- anti-CD20
Fc-Receptor protein binding assay
Rat line
FUT-/-s
Complement-mediated cell toxicity is the same
for FUT8-/- and Rituxan(not T-cell mediated ADCC)
Rituxan commercial product, 98 fucosylated
13
Very laborious, but apparently a big
payoff. Better selection? Why not use the
fluorescent LCA to select for the FUT8 KOs along
with G418 resistance (double, sequential
selection) ? Avoid 100,000 PCRs. . .
14
Second generation protein therapeutics
improvements over nature.
Or Isolation of recombinant protein mutants
with altered binding properties. Here Tissue
plasminogen activator (TPA, clot-buster)
Why are they doing this? Problem Binding of TPA
to liver cells leads to clearance from the
bloodstream Want to avoid clearance in TPA
therapy (anti-thrombolytic, clot buster) Know
MAb387 blocks binding to cultured hepatoma cells
(liver-like) Know MAb387 decreases clearance
rate. Goal Mutate the cloned TPA gene. Mutate
it in the MAb387-binding region ? mutant TPA
that 1) is hepatocyte binding-negative
(select) 2) is still protease (remains
catalytically active) (screen)
15
How could one do this? Select? Need to
characterize many mutant proteins, and find the
protein with the desired characteristic, and
then rescue the gene for that protein.
Mammalian cell transfectants But TPA is
secreted, so protein becomes divorced from the
DNA that coded it. My editorializing But
Coffino and Scharff had a technique for looking
at secretory variants (of myeloma cells)
Immunoprecipitate secreted proteins around
colonies grown in agar (Ig secretion,
precipitation by goat anti-mouse-Ig antibody)
16
All in soft agar
Medium in agar
Imagine Antibody in top layer MAb387 Colonies
CHO cell permanent co-transfectants of mutant
library TPA
Longer. Colonies may not make enough. But you
dont need a FACS ()
Precipitate -
Precipitate
Coffino and Scharff (Proc Natl Acad Sci U S A.
1971 Jan68(1)219-23.)
17
Consider an alternative
Phage display a way to link the variant protein
to its coding DNA Mutagenize the gene as a fusion
protein to a phage coat protein and make a
library in bacteriophage. The mutants will be
displayed on the surface of the phage and can be
panned for (or against).
DNA
DNA
DNA
Here, one would collect the members of the phage
library of mutant TPAsthat dont stick to
hepatoma cells, or to immobilized MAb387. But
lots of noise in a negative selection
non-stickers could be for many uninteresting
reasons (denatured, statistical, etc. )
18
Authors here use a mammalian cells as the carrier
of the DNA and the cell surface as a display
site. Via making a fusion protein to a membrane
anchor protein, DAF (peptide, really). What did
they do? Mutagenesis What region? 333 bp K1
(kringle-1), known to bind the MAb387, which
competes for hepatocyte binding (so assuming it
is the same target epitope). How did they get it
mutated? Error-prone PCR How did they isolate
just the kringle 1 region? PCR fragment. How
did they get the mutagenized fragment back
in? Introduced restriction sites at the ends, w/o
affecting the coding.
19
What did they put the mutagenized fragment
into? DAF TPA fusion protein geneHow did they
get it into into cells? Electroporation How many
copies per cell. And why is that important? One,
by electroporation at low DNA concentration. In
a transient transfection! Binding is dominant.
Lack of binding is recessive. How did they
select cells making MAb387-non-binding TPA? FACS
20
Sort the cells with low fluorescence
For reiteration of the process
21
How did they recover the plasmid carrying the
mutant TPA gene from the selected cells? Hirt
extraction Like a plasmid prep, lyse cells
gently, high MW DNA entangles and forms a
clot. Centrifuge. Chromosomal DNA ? soft
pellet plasmid DNA circles stay in supernatant.
Then re-transfect, re-sort in FACS. After 2
sorting rounds, test individual E. coli clones
60 are binding-negative.
22
MAb to protease domain
enriched
Collect these
No good
good
good
good
good
Low kringle-1 reactivity
MAb to kringle-1 domain
PE phycoerythrin (fluorescent protein)
23
AND
FACS selection can also work for an internal
protein
absence of
Predicted. freq. of dhfr- mutants 10-4
Still only 1 in 10 were mutants.
24
Hepatoma cell binding. How? Clone mutated
regions into regular TPA gene for testing (no
DAF, protein secreted) Label WT TPA with
fluorescein (FITC) (conjugated chemically) Mix
with hepatoma cells and analyze on a flow
cytometer (FACS w/o the sorter part). See
specific and non-specific binding. Subtract
non-specific binding the amount not competed by
excess un-labeled wt TPA.
25
Hepatoma cell binding assay measure comptetion
for binding of fluorescently labeled WT TPA
No competitor
WT
Compete. So still bind.
26
(No Transcript)
27
(No Transcript)