A chemiluminescence-based gel assay may be used as an indicator
of fermenter health. The assay is performed using a sample of fermentation
medium essentially directly from the fermenter (with no need for
complicated purification of the sample before performing the assay).
Results from the assay may be correlated with the final product
purification yield of a given protein/purification protocol system.
This may then be used as a predictor of purification yield before
actually performing the purification.
What is claimed is:
1. A method of predicting the yield of a protein product of a purification
process, wherein the protein product is obtained from cells in a
fermentation medium in a fermenter system, the method comprising
the steps of
a) obtaining multiple batches of fermentation medium containing
b) subjecting individual batches from step a) to the purification
c) measuring the percentage yield of the protein product of the
purification process for each individual batch of step b),
d) determining a quality rating from each individual batch of step
a), wherein the quality rating is a numerical value assigned to
an estimate of the percentage of desired protein product represented
in the total number of barnds observed on a Western blot,
e) determining the correlation between the yields measured in step
c) and the quality ratings determined in step d),
f) obtaining a test sample of fermentation medium from the fermenter
g) determining the quality rating from the test sample, and
h) using the correlation of step e) and the quality rating in step
g) to predict the yield of the protein product of the purification
2. The method of claim 1, wherein the Western blot assay includes
a chemiluminescent detection system.
3. The method of claim 1, wherein the product is recombinant factor
4. The method of claim 3, wherein the purification process comprises
the steps of
a) contacting the factor VIII with an anion exchanger under conditions
which allow the recovery of a fraction containing ion-exchange-purified
factor VIII, and
b) contacting the factor VIII with an immunoaffinity adsorbent
under conditions which allow the recovery of a fraction containing
affinity-purified factor VIII.
5. The method of claim 1, wherein the product is IL-2.
BACKGROUND OF THE INVENTION
The present invention relates generally to the production of target
proteins in cell culture, and specifically relates to a method of
assaying a sample of the cell culture medium to evaluate the quality
of the target protein in the sample.
Blotting procedures have been cited in the literature since 1975
when Southern published his method of DNA fragment identification
after transfer of the DNA from a gel to a nitrocellulose membrane
(Southern, 1975). This type of macromolecular transfer was followed
by the transfer of RNA onto a filter matrix, which was termed Northern
blotting (Alwine et al., 1977). The transfer of proteins from gels
onto membranes occurred in 1979 and was termed Western blotting
(Towbin et al., 1979). Since then, numerous reviews have been written
describing protein blotting (Dunbar, 1994; Gershoni, 1988). The
immobilization and detection of proteins generally involves five
basic steps. The first step is the electrophoresis of the proteins
on a PAGE-gel. Next, the proteins are transferred from the gel and
immobilized onto a membrane. Third, non-specific sites on the membrane
are blocked so as to increase the signal to noise ratio. The fourth
step includes the binding of specific antibodies to the immobilized
proteins and subsequent binding of a secondary antibody that recognizes
the primary antibody. Finally, the secondary antibody is detected
via a detection system, typically involving an enzymatic reaction.
The four main detection systems used on immunoblots are radiometry,
calorimetric, bioluminescence, and chemiluminescence. Radiometry
involves labelling the samples/antibodies with radioisotopes and
exposing the blot to autoradiography film. Although this procedure
is sensitive it requires the handling and disposal of radioisotopes
and a radioactivity safe facility. Furthermore, the exposure time
to the X-ray film is much longer than that for other detection methods.
Colorimetric detection methods involve using a secondary antibody
that is conjugated to an enzyme, for example alkaline phosphatase,
which will react with a colored substrate such as bromochloroindolyl
phosphate and nitroblue tetrazolium, to produce a color wherever
the antigen/primary antibody complex has reacted with the secondary
antibody. The colorimetric method is more sensitive and faster than
the radioactive method but does not give a permanent hard copy and
is not as sensitive as chemiluminescence or bioluminescence.
More recently, detection systems that have focused on the detection
of light have become more common because of their high sensitivity
and prolonged and rapid signal output. Bioluminescence and chemiluminescence
are the two most sensitive detection methods used for Western blotting).
Although both involve the emission of light, they primarily differ
in the substrate they use. Bioluminescence substrates such as luciferin
are natural products while chemiluminescent substrates such as luminol
are made synthetically. Bioluminescence detection involves the release
of activated luciferin from luciferin-o-.beta.-galactoside by .beta.-galactosidase
during its interaction with an alkaline phosphatase conjugated secondary
antibody. The luciferin is oxidized to oxyluciferin by luciferase,
with light being a product of the reaction. This technique has allowed
the detection of as little as 5 fg of protein (Geiger, 1994).
Chemiluminescence involutes the oxidation of a peracid salt by
horseradish peroxidase (HRP) that is conjugated to a secondary antibody.
This oxidation reaction raises the oxidation state of the HRP heme
group. As the electron attempts to come back to its ground state
it reacts with lumincil to form a luminol radical. As the luminol
radical decays it emits light, which is then detected on an autoradiography
film. The use of enhancers can prolong the luminol decay for up
to 24 hours, which is the main advantage of chemiluminescence over
bioluminescence. Chemiluminescence is as sensitive and rapid (often
only a few seconds of film exposure is necessary) as bioluminescence.
The sensitivity of the chemiluminescence can be significantly improved
if the secondary antibody has been conjugated with biotin. This
complex can then be exposed to avidin conjugated HRP conjugate and
reacted with the luminol reagents. Chemiluminescence is a preferred
method of detection because of its high sensitivity, its rapid speed
of detection, its prolonged emission time as well as the ready availability
of reagents in a kit format from a large number of suppliers.
Common applications of Western blotting include immunodetection
of antigenic sites on polypeptides and for amino acid sequencing.
Other applications for Western blots include the detection and characterization
of glycoprotein carbohydrate chains (Sato et al., 1998) and detection
of receptors zand the study of protein-protein interactions. This
has all been made possible by the extensive and sensitive detection
methods used in immunoblotting. Most applications, however, have
been confined to confirming the presence or absence of product in
cell extracts and not for assessing the overall quality of the product
for the prediction of downstream purification yields. (Kennel et
SUMMARY OF THE INVENTION
A sensitive, generic, electrophoretic based method has now been
developed for assessing the quality and the potential for purification
of target proteins directly from fermenter harvests, without any
need for concentrating the sample. This has been demonstrated with
recombinant Factor VIII (rFVIII) produced by a BHK (baby hamster
kidney) cell line.
In a preferred embodiment the technique involves running a desalted
sample of growth medium on SDS-PAGE followed by Western blot assay
using antibodies which bind the target protein, e.g. anti-Factor
VIII monoclonal antibodies. This technique provides a significant
improvement in sensitivity over conventional Western blot detection
techniques. A similar approach has been used previously to characterize
proteolytic fragments of rFVIII during cell culture (Kaufmnan et
al., 1988). Routine detection as low as 4 ng of target protein has
been accomplished by utilizing chemiluminescent substrates in combination
with an amplification system having an anti-mouse biotinylated secondary
antibody that reacts with avidin-labelled horse-radish peroxidase.
Furthermore, this assay has a high throughput and therefore is particularly
useful during fermentation development for rapid screening of multiple
fermenter conditions and ultimately for determining the potential
for purification. The assay has also been applied to the successful
prediction of purification process yield.
We have quite surprisingly found that the chemiluminescent Western
blot method (ZAP method) correlates well with other assays which
can be more difficult and time consuming to perform. The ZAP method
described herein is thus particularly useful as an alternative to
those assays (since the only method of monitoring rFVIII in the
fermenter is by activity titer). The ZAP method can be correlated
with purification yields and can be used to demonstrate that the
usual method of monitoring rFVIII (activity titer) is not predictive
of product quality. It is not the purpose of this report to assign
a cause for any of the examples indicated below but rather to demonstrate
that the ZAP method can be a sensitive tool under a variety of conditions.
A qualitative classification system was designed and used in conjunction
with the ZAP method to categorize the quality of a fermenter and
its harvest for monitoring purposes. Furthermore, the ZAP method
can be used in conjunction with the qualitative classification system
to determine the potential of the harvest for purification processing,
without having to actually carry out the purification on each sample.
The purification process can be any optimized series of steps to
provide a substantially pure product, i.e., particularly free of
cellular contaminants, preferably resulting in a product which is
at least 60% pure, preferably at least 80% pure, more preferably
at least 90% pure, still more preferably at least 95% pure, or most
preferably at least 98% pure. The steps may include standard purification
procedures well known in the art (see, generally, Scopes (1987)
and Deutscher (1990)), with selection of individual steps depending
on what other components are in the sample. Standard purification
methods include electrophoretic, molecular, immunological and chromatographic
techniques, including ammonium sulfate precipitation, density gradient
centrifugation, solvent extraction, gel electrophoresis, ion exchange
chromatography, size exclusion chromatography, hydrophobic interaction
chromatography, immobilized metal affinity chromatography (IMAC),
affinity chromatography, reversed-phase HPLC chromatography, and
chromatofocusing. Ultrafiltration and diafiltration techniques,
in conjunction with protein concentration, are also useful.
FIG. 1 schematically represents the structure of rFVIII. Recombinant
FVIII is a glycoprotein with an approximate molecular weight of
290 kD and includes multiple domains. In the intact rFVIII molecule
the A1 domain (at the N-terminus), the A2 domain, and the B domain
are all linked as a single polypeptide chain of an accumulated molecular
weight of about 210 kD. The 210 kD polypeptide chain is linked via
the A1 domain by a metal ion to a smaller polypeptide chain that
is about 80 kD. The 80 kD domain contains a C2 domain at the C-terminus,
a C1 domain, and an A3 domain. Recombinant FVIII from mammalian
cells has been shown to have a high level of structural heterogeneity
caused by different levels of glycosylation and intracellular processing.
In FIG. 1 the potential glycosylation sites are indicated by the
triangles. These different processed products will have varying
molecular weights and as such can be separated on an SDS-PAGE gel
and subsequently be detected on the Western blot.
A high degree of processing suggests that there may be problems
in clearing these products during downstream purification. It may
also imply that these species are by-products of extracellular proteolysis
that occurred once the target molecule had been secreted into the
medium. Another possibility is that the cells are being stressed
by a certain fermentation parameter so that they either produce
large amounts of extracellular proteases, which clip the target
molecule, or that they initiate an unacceptably high degree of internal
processing. The damaged product and low productivity fermentations
resulting from any of these scenarios is not always detected by
the usual activity assays (e.g. coagulation titer of FVIII). The
ZAP method provides another analytical tool for assessing the quality
of the product and the productivity of a fermentation.
BRIEF DESCRIPTION OF THE FIGURES
FIG. 1 schematically illustrates the structure of factor VIII.
Potential glycosylation sites are indicated by triangles. Amino
acid position is numbered conventionally, starting with the N-terminus
and ending at the C-terminus.
FIG. 2 is a graph showing the correlation of overall yield vs.
ZAP quality rating.
Materials and Methods
Desalting of Samples:
A detailed list of desalting reagents is given in Table 1. Frozen
samples (stored at -70.degree. C.) in either 15 or 50-ml centrifuge
tubes were thawed in room temperature water for not more than 10-15
minutes, vortexed, then placed on ice. During the thaw period, an
appropriate number of Econo-Pac.TM. disposable desalting (10 ml)
chromatography columns were placed in a PolyColumn.TM. rack. Each
sample was desalted in a separate (labeled) column. The column tops
and plugs were removed and the liquid drained into the rack buffer
tray. Each column was filled to the top and equilibrated with approximately
20 ml of 1.times.dialysis buffer. After the column stopped dripping,
exactly 3 ml of thawed sample were pipetted into the column and
allowed to migrate onto the resin. When the dripping stopped, labeled
15 ml (decapped) centrifuge tubes were placed under each respective
column. Exactly 4 ml of 1.times.dialysis buffer were pipetted into
the column and allowed to migrate into the resin. Samples of 4 ml
were collected in each tube. After the dripping stopped, the tubes
were capped, vortexed, and placed on ice. (Samples were stored at
-70.degree. C. for later analysis.)
TABLE 1 Desalting Reagents and Supplies 1) Econo-Pac .TM. 10 DG
Disposable Desalting (30 .times. 10 ml) Columns (Bio-Rad, Hercules,
CA) 2) PolyColumn .TM. Rack (Bio-Rad) 3) FALCON brand BlueMax .TM.
Disposable polystyrene 15-ml centrifuge tubes with caps (VWR, Bridgeport,
NJ) 4) 10X Dialysis Buffer, pH 7.4 (0.5 M TRIS, 1.5 M NaCl, 25 mM
CaCl2, 0.1% Tween 20) 5) 1X Dialysis buffer
A detailed list of SDS-PAGE reagents is given in Table 2. Samples
were prepared for SDS-PAGE as soon as possible after desalting.
Usually 33 .mu.l of each sample was pipetted into a corresponding
(labeled) microcentrifuge tube. To each tube 5 .mu.l of 10.times.
reducing agent and 12 .mu.l of NU-PAGE 4.times.LDS sample buffer
were added to make 50 .mu.l of total sample volume. The samples
were vortexed and then heated at 70.degree. C. to 100.degree. C.
for 2-10 minutes in a heating block. (Water was added to each block
well to insure uniform heating of the tubes.) The well location
was marked on one or two pre-cast 4-12% gels with a (Sharpie.TM.)
pen and the gels were placed into an Xcell II mini-cell SDS-PAGE
box. Approximately 200 ml of 1.times.MOPS buffer (1.times.) containing
500 .mu.l anti-oxidant was placed into the top reservoir of the
gel box. Samples, molecular weight markers, and rFVIII standard
were loaded into wells. When all samples were loaded into lanes,
the bottom buffer well was filled approximately 3/4 full with 1.times.X
MOPS buffer (without anti-oxidant). The gels were run at 200 volts
until the blue dye front reached the bottom and started to migrate
out of the bottom opening (approximately 45-60 minutes). While the
SDS-PAGE was running, the Western blot apparatus was set up to receive
the gels. A plastic tray was filled with approximately 200-300 ml
of 1.times.transfer buffer. For each gel being run on SDS-PAGE,
2 transfer sponges (part o the mini-transblot cell transfer apparatus),
2 mini-transblot filter papers, and 1 nitrocellulose paper were
placed in the tray and any air bubbles were removed. When the SDS-PAGE
gels were finished running, the gels were removed from the box.
TABLE 2 SDS-PAGE Reagents and Supplies 1) Xcell II Mini-cell SDS-PAGE
box 2) NUPAGE .TM. 4-12% Bis-TRIS (pre-cast) gels, 1.0 mm with 12
wells 3) NuPAGE .TM. MOPS SDS (20X) Running buffer 4) NuPAGE .TM.
(10X) anti-oxidant 5) NuPAGE .TM. (10X) Sample Reducing Agent 6)
NuPAGE .TM. (4X) LDS Sample Buffer 7) Molecular Weight markers:
SeeBlue Pre-Stained Standard and MultiMark Multi- Colored Standard
8) Heating Block, model 2002 with 3 module blocks, model 2069 (Lab-Line)
9) Microcentrifuge tubes-(0.5 ml or 1.7 ml) 10) rFVIII positive
control standard; diluted to a final concentration of 0.01U/.mu.l
using 10X Sample Reducing Agent, and 4X LDS Sample Buffer at the
appropriate volumes. This solution is heated to 70.degree. C. for
5 minutes, then stored at -70.degree. C. as 20-25 .mu.l aliquots.
Each aliquot is used once (i.e. thawed aliquots are not refrozen).
Reagents and supplies 1-7 are purchased from NOVEX (San Diego, CA).
A detailed list of Western blot analysis reagents is given in Table
3. Immediately upon removal of the gels from the electrophoresis
box, they were placed in the plastic tray containing the 1.times.transfer
buffer and all of the materials listed in the SDS-PAGE procedure.
A transfer "sandwich" was created as follows (use gloves):
A sponge was first placed at the bottom of a tray so that it was
covered with transfer buffer. A filter paper and then the nitrocellulose
membrane were placed on top of the sponge. The gel was placed carefully
onto the membrane so as to prevent tearing and to recognize its
orientation, followed by the second filter paper and the second
sponge. This entire "sandwich" was very carefully placed
inside the plastic transfer cassette, which was then placed inside
the transfer box. The transfer was runt overnight at 25-50 Volts.
Typically, 35 volts for 16 hours gave good results. After transfer,
the membrane was removed from the sandwich and placed in blocking
buffer for 60 minutes at 37.degree. C. with mild shaking on the
lab rotator. Hereafter, all antibody incubations and wash steps
were at 37.degree. C. with mild shaking on the lab rotator. The
membrane was washed in 50-100 ml of 1.times.TBST, 4 times for 5
minutes each. The primary antibody dilution factors were determined
experimentally to compensate for the titer of a specific lot. Typically,
the concentrations used were: R8B12 was at 107.5 .mu.g in 20 ml
(1.times.) TBST and C7F7 was at 231 .mu.g in 20 ml (1.times.) TBST.
The primary antibodies were either used alone or in combinations
at the same concentration listed. The membrane was incubated sequentially
with primary (1.degree.) antibody (at the appropriate dilution),
a 1:100,000 dilution of goat anti-mouse biotinylated secondary (2.degree.)
antibody, and a 1:500,000 dilution of HRPase-conjugated NeutrAvidin.TM..
Each incubation lasted for 1 hour and was followed by a washing
step. The membrane was incubated in 6-10 ml total volume of a 1:1
mixture of the SuperSignal West Femto Maximum Sensitivity Substrate.TM.
chemiluminescent reagents for 1 minute (at room temp) with mild
shaking by hand. Saran wrap was folded over the membrane so as to
avoid bubbles between the wrap and the membrane and excess wrap
was cut away with a scissors. The membrane was exposed to film in
a film cassette for a measured amount of time. Film exposure times
were adjusted as needed to clearly detect bands; typically, between
1 and 30 seconds was sufficient.
TABLE 3 Western Blot Reagents and Supplies 1) Mini-TransBlot Cell
transfer box apparatus (Bio-Rad) 2) NuPAGE (20X) Transfer buffer
(NOVEX) (2 Liters of 1X Transfer buffer is made as follows: 100
ml 20X Transfer buffer, 200 ml A.C.S. methanol, 2 ml 10X anti-oxidant,
1698 ml water) 3) Nitrocellulose paper, 0.2 .mu.m (8 .times. 8 cm)
(Pierce, Rockford, IL, USA) 4) Mini-TransBlot Filter Paper (Bio-Rad)
5) High Current Power Supply, model VWR570 (VWR) 6) Stir plate (VWR)
7) Plastic (324L .times. 260W .times. 70H mm) tray (VWR) 8) ISS
or Tupperware trays 9) Low Temperature/B.O.D. Incubator, model 2005
(VWR) 10) Lab-Line brand (platform) Lab Rotator, model 1314 (VWR)
11) Superblock Blocking Buffer in PBS (Pierce) 12) TRIS-buffered
Saline with Tween 20, or "TBST" (Made as a 4X concentrated
stock solution: 80 mM TRIS-base (pH 7.4), 2 M NaCl, 0.2% Tween 20)
13) Goat Anti-mouse ImmunoPure .TM. Biotinylated Secondary Antibody
(Pierce) (Diluted to 1:100,000 final dilution). 14) NeutrAvidin
.TM. Horseradish peroxidase-conjugated (HRP) (Pierce) (Diluted to
1:500,000 final dilution). 15) SuperSignal BLAZE .TM. Chemiluminescent
Substrate (Pierce) 16) Saran wrap (VWR) 17) Hyperfilm ECL High Performance
Chemiluminescent Film (Amersham Pharmacia Buckinghamshire UK) 18.)
Medical Film Processor, model QX-70 (Konica) 19.) Film Exposure
(8 .times. 10 in.) Cassette (Sigma, St. Louis, MO) with 2 DuPont
Cronex .TM. "Lightening Plus" enhancer screens
This procedure is designed as an ultra-sensitive analytical method
for the detection of femtomolar amounts of a protein. The following
non-limiting examples demonstrate optimized analysis of rFVIII from
BHK cell fermenters and harvests.
Optimization of the Method in rFVIII Production:
The ZAP method was optimized for determination of the presence
of the 210 kD, the 80 kD and any other truncated or unprocessed
forms of rFVIII which exhibit rFVIII-type antigenicity to the antibodies.
The 210 kD and 80 kD in appropriate ratios served as indicators
of intact rFVIII (and hence good fermenter health) while any other
fragments served as indicators of different degrees of product fragmentation
(and hence poor fermenter health). The combined indicators were
incorporated into a classification system that was used to determine
the quality of the product, the potential for providing high yields
from the purification train, as well as the general health state
of the fermenter (see Table 5 for classification system).
Many aspects of the ZAP technique were investigated for optimization
purposes. These elements were studied in se-quence beginning with
the optimization of the sample preparation and including the desalting
step, the type of SDS-PAGE gel, the type of blotting paper, the
type of blocking buffer, optimizing primary and secondary antibody
concentrations, optimizing the avidin concentrations, evaluating
vendor chemiluminescent reagents. Although this technique has many
variables, not all were optimized since the variables that were
optimized resulted in a method that provided satisfactory results.
Initially, carrying out an acetone precipitation after the desalting
of the sample was necessary to see strong bands on the ZAP since
the ZAP methodology had not been optimized for sensitivity. Once
the ZAP sensitivity was optimized the necessity of the acetone precipitation
step was re-evaluated. Eliminating the acetone precipitation step
from the ZAP methodology was viewed as critical since it would accelerate
the process by 3 to 4 hours and also improve the reproducibility
of the technique.
One problem with carrying out acetone precipitation is that all
traces of acetone need to be removed prior to SDS-PAGE. In some
cases, especially when there is a large sample number, there may
be a sample in which the acetone has not been fully evaporated.
Failing to evaporate a significant portion of the acetone was observed
to result in smearing of the sample on the gel. In this case, it
was impossible to obtain any information about the sample.
Carrying out the ZAP without having to do an acetone precipitation
improved accuracy, reproducibility, and quality of the results as
well as significantly reducing the time for the process.
It was found that including 0.01% Tween 20 into the sample prior
to desalting led to signal enhancement, apparently by preventing
non-specific interactions with the desalting column. In practice,
the range of Tween (or any other non-ionic surfactant) may need
to be adjusted to yield optimum results, for example between about
0.001% and about 1.0%, or more preferably between about 0 01% and
Selection of SDS-PAGE Gels:
SDS-PAGE gels were examined from two different vendors to observe
their impact primarily on protein transfer. Gels from Novex and
FMC (Roskland, Me.) were compared. The gels from both vendors were
developed by chemiluminescence as described in the materials and
methods section. The Novex gel had a lower background than the FMC
despite both gels being run and developed under identical conditions.
It was also noted that the FMC gel was more difficult to manipulate
during transfer and was very brittle. As a result, the FMC gel fell
apart during our handling. We concluded that Novex gels would be
preferred, though other gel products commercially available could
potentially be used in the assay, also.
Nitrocellulose and polyvinyldivinyl fluoride (PVDF) membranes were
compared for carrying out transfers from SDS-PAGE gels. A significant
improvement of signal to background ratio was observed when using
nitrocellulose. Such a result is to be expected since, because of
their hydrophobic nature, PVDF membranes tend to bind more proteins
than nitrocellulose, which is a relatively more inert membrane.
Blocking Buffer Formulation:
Blocking buffers were used to quench any non-specific binding onto
the membrane thereby reducing background staining. Some of the common
blocking reagents include bovine serum albumin, gelatin, Tween 20
and a variety of non-fat milk products. We tested bovine serum albumin,
non-fat dry milk (from a variety of sources), as well as a commercially
available blocking agent of unspecified constituency from Pierce,
known as SuperBlock.TM.. Tween 20 (0.05% v/v) was in all blocking
solutions tested except for SuperBlock which was tested at 0.01%,
0.05% and 0.2% v/v Tween 20 solutions.
The blocking buffer solutions were evaluated based on their ability
to provide a low background as judged by the absence of the molecular
weight markers and a lightening of the background. The inability
to quench the signal from the molecular weight markers was an indication
that the blocking buffers still allowed some cross-reactivity to
occur between the primary antibodies and the molecular weight markers.
This would indicate that the buffers may then also not be sufficiently
blocking BHK cell antigens from the fermenter that the primary antibodies
may cross-react with. The 10% non-fat dry milk from BioRad was found
to be superior to the other blocking buffers tested since it produced
very low background and was able to quench all of the molecular
weight markers. As a result, we chose 10% non-fat dry milk from
BioRad as our preferred blocking buffer.
It must be noted that this is only the blocking buffer of choice
for this specific system. It may be necessary to optimize the blocking
buffer for the particular protein being investigated; the optimization
may be accomplished by methods well known in the art and while generally
following the examples contained herein, varying process parameters
to optimize the desired outcome. As a cautionary note, blocking
buffers containing milk products should be investigated carefully
if lectins or antibodies that recognize carbohydrate moieties are
to be used in the assay, since milk contains large amounts of sugar
that may block binding. If this is the case, then we would recommend
the use of the SuperBlock buffer with 0.2% Tween.
Selecting Primary Antibodies:
A number of monoclonal antibodies were examined for use on ZAPs
as primary antibodies. These were 19A9 (against the B-domain), 58.12
(against the 90 kD N-terminus), C7F7 (against the N-terminus of
the light chain), R8B 12 (against the 90 kD) ESH-4 (light chain),
ESH-5(light chain), and ESH-8 (light chain). A polyclonal was also
tried without much success and is not represented here. Furthermore,
the use of a polyclonal would not be as informative as using a monoclonal
antibody due to its intrinsic loss of specificity and so was not
further pursued. Results are described in Table 4.
Based on the above results, we selected the R8B12 and the C7F7
antibodies as the main primary monoclonal antibodies to be used
during Western blot development. Other antibodies such as the ESH
antibodies could be used but were not chosen since significantly
larger concentrations of these antibodies would need to be used
to obtain the equivalent signal as the C7F7 and R8B12 which are
available in-house. These antibodies provide us with sufficient
information pertaining to the presence or absence of 210 and 80
kD as well as a sufficient number of fragments to inform us of the
fermentation quality. We do not need to see every fragment that
is; produced in the fermentation since the ones that can be detected
by the C7F7 and R8B12, can act as good indicators. Furthermore,
too many bands on the ZAP will detract from being able to accurately
quantitate the more important 210 kD and 80 kD bands. Other monoclonal
antibodies which bind with specificity to rFVIII may be used to
yield comparable results, particularly if the antibodies are selected
to yield complementary information on the species in the sample.
Preferably, more than one antibody is selected, such as two or three
different antibodies, with four, five, six, or more different antibodies
being selected in preferred embodiments.
TABLE 4 Attributes of antibodies tested Monoclonal Antibody Attribute
Result 19A9 Directed against the B-domain Positive but too many
bands- Not chosen 58.12 Directed against 90kD portion Negative,
too faint-Not chosen of 210kD fragment C7F7 Directed against N-terminus
of Positive, 80kD clearly observed; the 80kD not antigenic to BHK
proteins- Chosen R8B12 Directed against 90kD portion Positive, 210kD
clearly of 210kD fragment observed; not antigenic to BHK proteins-
Chosen ESH-4 Directed against 80kD Negative; No bands observed.
Not chosen ESH-5 Directed against 80kD Negative; No bands observed.
Not chosen ESH-8 Directed against 80kD Negative; No bands observed.
The approach one could use to select the primary antibodies for
other target molecules is essentially similar to above. The approach
can be broken down to two components: Firstly, an understanding
of the target molecule's SDS-PAGE profile needs to be known. For
example, is the molecule monomeric or multimeric under the chosen
SDS-PAGE conditions or are there degradation products or aggregates?
Also, determining if the target molecule has isoforms which serve
as antigenic sites is critical in choosing the primary antibody.
Detection of isoforms can be carried out easily on a 2D-gel with
subsequent Western blotting. After the antibodies are chosen on
the basis of their target molecule antigenicity, the second step
in the process is to ensure that the successful antibodies provide
a high "signal to background" ratio. This is done by carrying
out a titration of the antibodies in a preoptimized blocking buffer.
Along with background reduction, non-specific interaction with the
host cell proteins should also be demonstrated. Using such a procedure
we have been able to use the ZAP for determining the quality of
Selecting Chemiluminescent Reagents:
Chemiluminescent kits were examined for improving the signal to
noise ratio. The various kits tested were ECL (Arrersham, Piscataway
N.J.), Renaissance (NEN, Boston, Mass.) and BLAZE (Pierce, Rockford,
Ill.). All kits were tested as per manufacturer's protocol. The
BLAZE reagent was by far the most sensitive and also provided the
lowest level of background. The manufacturer claims that it can
detect down to femtomole levels of a single protein, and we could
routinely detect 7.3.times.10.sup.-15 moles of rFVIII per well.
Other chemiluminescence kits could be used but would need to be
optimized for time of exposure and concentration of samples.
The secondary biotinylated antibody supplied with the Pierce kit
was titrated over the range of 10,000 to 100,000 against the avidin
tagged horseradish peroxidase antibody from 100,000 and 500,000
dilution. The ratio of 100,000 dilution of the secondary to 500,000
dilution of the NeutrAvidin.TM. worked best.
Qualitative Classification System for Determining Quality Rating:
Samples loaded on an SDS-PAGE gel were usually denatured by the
addition of an excess amount of SDS and a reducing agent and then
heating at 70.degree. C. to 100.degree. C. for 2 to 5 minutes. Under
denaturing conditions the metal ion holding the two rFVIII polypeptides
together is dissociated leaving free 210 kD polypeptides and free
80 kD polypeptides. The 80 kD was observed to run as a doublet under
normal SDS-PAGE running conditions. These were the main species
observed, and if this was the case the quality rating of the product
was considered excellent. Other species that were observed included
a 90 kD (the A1 and A2 polypeptides processed from the B-domain)
and any other combination of molecular weight due to protein processing.
Free 80 kD was also observed. The intensity ratio between the 210
kD and 80 kD can serve as an indicator of free 80 kD (e.g. if the
80 kD is more intense than the 210 kD) and as such plays a role
in the established qualitative classification scheme outline below.
Based on this information, the following qualitative classification
system for determining quality rating was established for fermenters
and their harvest.
TABLE 5 Qualitative ZAP Classification System (rFVIII) Quality
Rating (number) Description Excellent or 4 Strong 210 kD and 80
kD. No fragments under 80 kD or between 210 kD and 80 kD. No bands
or smearing above 210 kD. Proceed with purification. Good or 3 It
has a strong 210 kD band and 80 kD band. Has fragments either above
the 210 kD or below the 80 kD or between the 210 kD and 80 kD, but
not in all areas. May have some low level smearing around and above
the 210 kD. Proceed with purification. Fair or 2 Has a weakening
210 kD and/or 80 kD such that the ratios are different to the standard.
Has fragments in more than one zone (ie. <80 kD,> 210 kD or
210 kD to 80kD and has much smearing. If purified will give rFVIII
but perhaps lower yield or lower purity may result. Bad or 1 Missing
either the 210 kD, 80 kD or both. Should not be purified.
Guidelines for a classification system for determining quality
rating for a general target protein are given in Table 6.
TABLE 6 Qualitative ZAP Classification System (general target)
Quality Rating (number) Description Excellent or 4 Target molecule
in its monomeric form is present with no detection of any other
antigenic pieces of the target molecule. Target molecule accounts
for nearly 100% of the bands seen on the ZAP Good or 3 Target molecule
in its monomeric form represents majority of the bands seen on the
ZAP i.e. in excess of 50% but less than 100% Fair or 2 Target molecule
in its monomeric form represents equal or less than the majority
of the bands seen on the ZAP i.e. .ltoreq.50% Bad or 1 Target molecule
in its monomeric form is either not detectable or not indicative
of an intact molecule (e.g., if target is a multimer, such as rFVIII,
in the instance one of the target molecule's subunits is missing.)
Analyzing Fermenter Growth Conditions:
To demonstrate the general utility of the ZAP assay, the technique
was applied to samples from rFVIII producing cells grown under different
conditions. Fermentation was conducted either in the presence or
absence of human plasma protein solution (HPPS). HPPS consists primarily
of albumin but also contains up to 30 other proteinaceous components
such as haptoglobin, transtherytin, ceruloplasmin and vitamin D
binding protein. HPPS was included in the fermentation medium at
2.25 g/L to help protect rFVIII from instability issues such as
proteolysis. Assuming a specific rFVIII activity of 5000 U/mg, an
average titer of 1.0 U/ml and that no proteins other than HPPS and
rFVIII are present, then rFVIII would represent approximately only
0.009% of the total protein content in these fermentations. This
percentage is most likely to be even less if we consider the protein
contribution made by BHK cell secretion. Such a poor rFVIII specific
activity made it impossible to visualize rFVIII from a fermenter
on an SDS-PAGE gel using any type of staining technique. Low concentrations
of rFVIII also made it difficult to detect rFVIII from a fermenter
using Western blotting with a chromophoric or radioisotope detection
The cell fermentation medium used in the fermentation runs conducted
in the absence of HPPS had essentially into added protein other
than recombinantly-derived insulin (10 mg/L). Insulin is .about.5
kD and should not interfere with the primary antibodies. In fact,
it may even run off the 4-12% SDS-PAGE gels. The only other proteins,
aside from rFVIII, that may have been detected were those secreted
by the BHK cells. In fact, an A.sub.280nm /A.sub.260nm reading of
fermenters routinely gave about 0.02 to 0.05 mg/ml level of protein,
assuming an extinction ci)efficient of 1.0 mg/ml.cm.sup.-1 for 1.0
absorbance unit and corrected for the presence of DNA. The rFVIII
titers were routinely about 1.0 U/ml, and so rFVIII represented
about 0.4% of the total protein content of harvests from the HPPS-free
fermentations. Nevertheless. rFVIII from HPPS-free fermentation
was not discernible from any impurities on an SDS-PAGE system with
any type of staining technique when the samples are taken directly
from the fermenter or harvest lines. Chromophoric detection systems
struggled to detect the 210 kD and 80 kD and certainly failed to
pick up minor fragments of rFVIII. Chemiluminescence detection was
therefore applied to analyze rFVIII from HPPS-free fermentation
taken from the fermenter and harvest lines.
Fermenter harvests from both HPPS-containing and HPPS-free fermentations
were examined on a Western blot using the chemiluminescence detection
method. By using the optimized chemiluminescent method the signal
from the 210 kD fragment of rFVIII from the HPPS-containing fermentation
matches up Oust as for HPPS-free rFVIII, below) with purified standard
rFVIII and is clearly observable. The 80 kD piece is masked by an
HPPS component speculated to be transferrin and so cannot be seen.
Nevertheless, any degradation that may be induced between the 210
kD and the 80 kD will be seen. The large amount of albumin in the
sample is detected non-specifically by the primary antibodies but
does not appear to interfere with the clear visualization of the
210 kD band. It may, however, mask rFVIII fragments that have a
similar size as the albumin The samples tested showed excellent
quality product and would be classified as class 4 type fermenters.
Several samples from HPPS-free fermentations taken from the fermenter
and harvest line were also tested. They all represented approximately
0.8 to 1.2 U/ml of rFVIII or 0.08 to 0.012 U/well. The 210 kD and
the 80 kD bands seen in the samples ran parallel to the bands from
the HPPS-containing sample. This observation, along with the strong
reactivity these bands had against the C7F7 and R8B12 monoclonal
antibodies, was used as confirmation that the 210 kD and 80 kD bands
in the samples were in fact the rFVIII bands. The minor fragments
observed in the samples, but not in the rFVIII, were further confirmed
to be pieces of rFVIII by probing rFVIII-free BHK (blank transfected)
cell harvest with C7F7 and R8B12 and not observing any detectable
bands. This experiment served as a negative control and did not
reveal any fragments. This observation further supported that the
fragments reacting with the C7F7 and R8B12 antibodies were truly
pieces of rFVIII and not proteins released by the BHK cells.
Coagulation Titer vs. ZAPs:
The experiments described in this example were designed to demonstrate
that coagulation titer is not always a good predictor of product
quality. A physical analysis of the products by ZAP should be used
in combination with coagulation titer for determining the properties
of rFVIII produced by the fermenter.
In this experiment, rFVIII samples from one particular fermentation
run were taken from the fermenter and harvest line and were analyzed
by ZAP. The samples had a titer of about 0.87 U/ml, which is considered
to be within the acceptable activity range and would normally be
processed for purification. The ZAP, however, clearly indicated
that although there is an acceptable titer, there is no 210 kD fragment
and only some 80 kD. Purification of harvest from this fermenter
using the automated purification train, followed by analysis on
an SDS-PAGE gel, also indicated the absence of the 210 kD fragment
and an overall recovery of only 26%.
These results clearly indicated that this fermentation run was
providing an unacceptable product. This is despite all other fermentation
parameters, including titer, being normal. Hence, without the ZAP,
this fermenter would have been considered to be operating normally
and be recommended for proceeding to purification. The outcome would
have been a poor yield and, a failed lot.
In the past, coagulation titers have been the main determinant
of fermenter-produced rFVIII quality. High coagulation titers were
interpreted as a good result while low titers were interpreted as
pocor. However, other species in solution may demonstrate rFVIII
activity, e.g. fragments of rFVIII or possibly aggregates. The B-domain
is not required for coagulation activity since during this particular
assay, thrombin is used to cleave the B-domain and obtain active
rFVIII (Factor VIIIa) (Pemberton et al., 1997). The A1 and A2 domains
only need to be held by electrostatic interactions, and as long
as they are metal-bridged to the C2.multidot.C1.multidot.A3 polypeptide,
then there will also be activity. It is also feasible that some
truncation of the A1 N-terminus and/or truncation of the A2 and/or
A3 C-terminus may still elicit activity. The net result may be disastrous
since it may then be interpreted that the fermenter is producing
good quality rFVIII, but by the end of the purification process
very little activity is recovered. This may be due to the smaller
activity producing fragments being purified away because of their
differential chromatographic retention. Such truncated forms would,
however, translate into a variety of molecular weights on SDS-PAGE
and hence be easily detected by the ZAP. By carrying out the ZAP
we will be able to verify the extent to which the full molecule
rFVIII contributes to coagulation titer in a sample.
Correlation Between ZAP Classification and Purification Yield from
Until now, it has been difficult to ascertain with any accuracy
the recovery of rFVIII from a purification process without carrying
out some rapid scaled-down purification process (such as the one
described below). In particular, the use of Western blots to predict
purification yields (and therefore purification process performance)
from an unprocessed fermenter sample has not previously been described
in the literature.
The complex nature of the purification process takes purification
yield prediction, directly from an unprocessed fermentation sample,
difficult. Purification yield is dependent on many parameters including,
but not limited to, the level and nature of the protein impurities,
the level and integrity of the target molecule and the ratio of
these values (i.e., the specific activity of the target molecule).
Other parameters include the number of purification steps and the
robustness of the process. So many complex variables makes an a
priori prediction of the purification yield difficult from a crude
fermenter sample. As such, a priori prediction of purification yields
using Western blots is novel.
Reducing or even eliminating any extraneously added proteins, such
as HPPS, from the fermentation reduces their impact on the purification
yield of the target protein. Furthermore, the number of purification
steps (where losses of the target protein may occur) necessary to
yield a purified product is reduced. This in turn should improve
the robustness of the purification process. Hence, by eliminating
the extraneous proteins we significantly improve the chances of
successfully using an unprocessed fermentation sample to predict
purification yield. This is certainly one of the rationales for
conducting the fermentation under HPPS-free conditions. The advantages
of being able to predict the purification yield from an unprocessed
fermentation sample are significant. For example, a manufacturing
run can be cut short, before any purification has been carried out
on a lot where the purification yield has been predicted to be uneconomical.
With such an advantage in mind, an attempt was made to correlate
the ZAP classification (as described above) with the rFVIII purification
yield obtained from fermenters under HPPS-free conditions.
The small-scale purification train consists of two sequential chromatography
steps. The first step uses ANX media, which is a highly crosslinked
agarose and is an anion exchange adsorbent from Amersham Pharmacia
(Piscataway, N.J.) and primarily binds rFVIII through the B-domain.
Binding to the B-domain is most likely through the negatively charged
sialic acid groups. The second step is an immunoaffinity step which
utilizes an immobilized monoclonal antibody that binds to the light
chain. As such, it will bind any whole or fragmented piece of rFVIII
that contains the N-terminus of the rFVIII light chain. Being an
affinity chromatography step it successfully removes the majority
of impurities in addition to any rFVIII that does riot have the
light chain attached. After these two columns, the rFVIII (derived
from HPPS-free fermentation) is usually at least 80% pure. While
ANX media and the immunoaffinity media are preferred materials used
in this purification protocal, other similar media are well known
to those skilled in the art and may be used with essentially equivalent
results, Where the protein of interest is not rFVIII, a separate
purification protocol will have to be designed and optimized to
suit the protein of interest. Such design and optimization is generally
within the capability of those skilled in the art of protein purification.
The purification yield data was obtained from the scaled-down version
of the two column anion exchange/immunoaffinity (ANX/C7F7) small-scale
purification trains. Percentage yield was calculated by dividing
the total number of coagulation units loaded onto the ANX column
by the total number of coagulation units recovered from the C7F7
column, and then multiplying by 100%. The same fermenters were analyzed
on ZAPs (one day earlier) and classified using the sample classification
system described in Table 5. A correlation graph was then draw between
ZAP classification and purification yield.
To establish a correlation. as used herein, is to establish a statistical
relationship between two sets of data (e.g. the assigned quality
ratings and the corresponding yields) by fitting the data to a linear
or curvilinear function and obtaining a best fit of the data to
the function. The graph in FIG. 2 indicates a correlation coefficient
of 0.81 with sample number n=21, if we assume a y=mx+c function.
A better fit might be attained if the best fit line is drawn such
that it tapers off at the class 1 level and shows more of a saturation
effect at class 4. Nevertheless, from class 3 and 4 to class 1,
it is observed that the higher the classification level the greater
the purification yield. This chart supports the concept that a correlation
does exist between purification yield and sample classification
such that an approximate prediction of purification yield may be
made that is based on ZAPs.
The anion exchange step and the antibody affinity step achieve
the majority of the purification and incur the majority of the yield
losses. Any additional columns that may be added downstream to the
purification process will be primarily for removal of specific impurities
such as DNA and/or viruses and it is not anticipated that there
will be more than 10 to 15% loss in these steps. This percentage
is well within the current error of the correlation graph when using
the qualitative classification system and so the impact on the correlation
would be minimal. It is not intended that this correlation chart
be used where the protein of interest is other than rFVIII. A new
correlation profile would need to be established for the protein
and its purification protocol. This will easily be accomplished
by following the process described herein of classifying ZAP results
and correlating them with the purification yields.
Despite the use of a qualitative classification for the ZAPs the
correlation between ZAPs and purification yield has been reasonable.
The ZAP should now not only be considered as a tool to detect the
general health of the fermenter, but also as a predictor of the
approximate, overall purification recovery for rFVIII. In fact,
such an approach would be amenable for any target molecule, especially
where the number of purification steps is small (preferably no more
than 3) and the cells are grown in a protein free fermentation medium.
Using the ZAP to Determine the Quality of IL-2 in Fermenters:
Interleukin-2 (IL-2) is a cytokine with molecular size of approximately
17 kD. It is expressed by a recombinant Chinese Hamster Ovary (CHO)
cell line which contains an expression vector coding for the IL-2
protein and which is grown in serum free culture fluid in the fermenter.
Analysis of purified IL-2 on a 2D gel electrophoresis system indicated
the presence of up to 17 different isoforms and the presence of
dimers as well as higher molecular weight multimers. The selection
of antibodies was carried out as described earlier. The antibodies
chosen were able to detect all IL-2 isoforms as well as the aggregated
forms of the IL-2. Observations are shown in Table 7.
TABLE 7 ATTRIBUTES OF ANTIBODIES TESTED Antibody Attribute Result
20-IR05 Directed against IL-2; Positive but too many bands; rabbit
polyclonal not antigenic to CHO proteins-Not chosen 10-172 Directed
against IL-2; Positive, all bands seen; not mouse monoclonal antigenic
to CHO proteins- Chosen AF-202-NA Directed Against IL-2; Signal
too weak- not chosen mouse monoclonal Monoclonal antibodies were
obtained from Fitzgerald Industries International Inc. #10-172.
Aggregates constitituted very low levels of the IL-2 produced by
the fermenters. However, they did represent a loss of IL-2 productivity
in the fermenter and so it was desirable to prevent them from forming
and preserving them in the monomeric forms. Hence a sensitive Western
blot was needed to be used to monitor the aggregates. By using the
ZAP for IL-2 detection in the fermenters the relative degree of
aggregation was able to be determined.
A generic chemiluminescence based assay system (denoted ZAP) to
detect antigens on Western blots has been optimized, and its application
as a predictor of protein recovery has been demonstrated. The ZAP
can be used to characterize the quality of dilute target molecules
directly from cell cultures without the need for any purification,
in a high throughput manner. Such a technique is ideally suitable
for a manufacturing environment where the ZAP can be used to prevent
the processing of poor quality product as well as provide data that
supports the processing of product where it has been previously
unclear if processing should proceed. In both cases, employment
of the ZAP would result in improvements in the cost of goods as
well as improvement in general efficiency. The ZAP is also ideally
suited for use during development of cell lines, fermentation development,
and purification process development.
This report describes the optimization of the ZAP to primarily
suit rFVIII, but with some minor adjustments it may be applied to
any protein, including bikunin, IL-4 and pigment epithelium derived
factor, and antibodies such as anti-IL-5. The protein may be obtained
from either a procaryotic or a eucaryotic cell growth medium. This
is dependent on the availability of antibodies that recognize relevant
epitopes of the target molecule. Although not all parameters were
examined during the ZAP optimization, the current ZAP protocol is
in a functionally optimized format.
The ZAP assay could also be performed using primary antibodies
that have been biotinylated. This would decrease the time the assay
takes by about 1.5 hours since the HRP conjugated-avidin could be
used directly in place of the current secondary antibody. This may
also result in an improvement in the signal to noise ratio through
the elimination of one of the steps. The use of other antibodies,
like 59.7 and 39MH8, which bind to rFVIII may reduce or eliminate
the smearing that is associated with the C7F7 primary antibody.
Removing the smearing becomes particularly important to enable the
ZAPs to be quantitative.
For rFVIII, titer (as measured by the coagulation or chromogenic
assay) has been the main method of assessing product quality during
fermentations. The optimized ZAP method adds to the existing arsenal
of probes that are used to monitor protein production in fermenters
by overcoming some of the weaknesses of the existing probes. For
example, the ZAP method was used to demonstrate that in some cases,
measurable titer might be obtained with truncated forms of rFVIII
that are pharmacokinetically unacceptable. Previously, without processing
the batch through the purification scheme, it would be impossible
to determine the existence of these truncated forms of rFVIII. As
a result, either scaled-down purification trains had to be run as
indicators of yield or only activity assays were used, and the impact
was sometimes reflected by poor yields during purification. By then,
however, it would be too late, with much money and time spent arriving
at an unsatisfactory conclusion. Using the ZAP method, these truncated
forms can be visualized prior to initiating any type of purification
process thus improving the efficiency within manufacturing.
The second indicator of product quality has been the yields of
the rFVIII after purification. A fully automated, two column, scaled-down
version of the initial steps in the purification process was developed
as an analytical technique. Although this technique gives an excellent
indication of the product yield and purity resulting from the purification
process, it has a poor throughput taking 20 hours per sample with
a maximum of 4 samples per week, per technician. Alternatively,
the gel-based chemiluminescence assay can be run in 8 hours with
a throughput of up to 100 samples per week, per technician. It provides
a visual characterization of rFVIII quality that can be independent
of activity. As such, it can be used to determine if any fermentation
should proceed to the purification stage. It can also give a general
indication of fermenter health by observing the extent of product
fragmentation, which may be a result of aberrant cell metabolism
or extracellular proteolysis.
Furthermore, the scaled-down purification does not give any indication
of general fermenter health other than rFVIII recovery. Indeed,
a Western blot and an SDS-PAGE gel still need to be run to obtain
information about purity and product quality. The scaled-down purification
version does not give an indication of truncated forms of rFVIII
that are produced because of some fermeriter variation that may
not ultimately affect purification. This, of course, is irrelevant
if we are only studying the effect of the fermentation variable
on purification. In some cases, however, such as in the development
of a cell line, purification is not always the end goal. The ZAP
method is particularly useful in these instances, since it will
differentiate truncated forms of rFVIII/proteins that may not have
an impact on the purification (i.e. they flow through the columns
or bind so tightly that they only come off during the regeneration
of the column).
Nevertheless, the ZAP used in conjunction with the quality rating
classification system (despite its qualitative rather than quantitative
nature) has shown excellent correlation with purification yield.
This raises the question of whether small scale purification trains
need to be used at all as a tool for revealing product quality or
even yield unless further analysis is required, such as a carbohydrate
map. The data presented in this report clearly support that ZAPs
could replace the need for purification to evaluate product quality.
The ZAPs cannot be used during development phases as an a priori
tool for predicting purification yield. In this scenario, the ZAP
would first require data input from the purification trains in order
to establish the classification criteria and to correlate with meaningful
The ZAP method may be extended to allow quantitative evaluation
of the antigenic bands. ZAP quantitation would improve the accuracy
of purification yield prediction by replacing the qualitative classification
system for quality rating currently used with a numerically quantitative
range. Furthermore, it would allow for a new correlation to be evaluated
between Western blot purity of the fermenter sample and final SDS-PAGE
silver stain purity of rFVIII, after purification. This would allow
quantitatively reliable purity predictions from ZAPs without carrying
out any purification. This is feasible since, at least for rFVIII
from HPPS-free fermentation, the final product will be greater than
80% pure, or more typically greater than about 90% or 95% pure,
with the remaining species most likely to be truncated forms of
rFVIII which have co-purified.
Another contribution that quantitation may allow is the ability
to determine the specific activity of rFVIII, not relative to other
(non-rFVIII) impurities, but rather to the level of active rFVIII,
as well as to the degree of rFVIII heterogeneity. Aspects of this
have already been noted in Example 3 where rFVIII standard and fermenter
samples which had equal titer showed differences in band intensity.
By using an acceptable standard with an accurately known titer,
a comparison of the relative intensities to the test samples may
be made for determination of the specific activity of rFVIII directly
from the fermenter. Determining the specific activity in this way
will enable such things as the estimation of contribution that rFVIII
heterogeneity has on purification losses, particularly during purification
Typically, quantitation is carried out by conventional densitometers
that scan the autoradiogram. Although this technique has been used
successfully for many years, it is dependent on the optimum performance
of the X-ray developing machine and developing solutions. More recently,
an Image Station.TM. from Eastman Kodak (Rochester, N.Y.) has been
introduced which is able to directly capture the light output from
the chemiluminescing Western blot without the need for exposure
onto an autoradiogram. With equivalent sensitivity, the Image Station
bypasses the developing machine and eliminates the need for autoradiography.
The Image Station is then capable of being used in a densitometry
mode where it can quantitate the light emitting bands on the Western
Other potential applications of the ZAP includes its use in the
characterization of the glycosylated component of rFVIII. In this
case, a lectin could be used to specifically identify the presence
or absence of sialic acid groups, directly from the fermenter and
without any purification. This would be beneficial in detecting
fermenter batches of rFVIII that may have been affected by sialidases.
The ZAP can be a sensitive tool under a variety of conditions.
The systematic optimization and development of key steps involved
in the Western blot procedure as well as the optimization of the
reagents used in the chemiluminescent detection system have been
described herein. Overall, the ZAP has been shown to be a sensitive
and generic method with a diverse range of applications. This technique
can be used to predict product quality and purification yields.
A classification system (see Table 5) for assigning a quality rating
was designed to categorize the quality of a fermenter and its harvest
for monitoring purposes. The classification system used for rFVIII
should be useful for other proteins of interest, too. See Table
6. Furthermore, the correlation can also be used to determine the
potential of the harvest for purification processing without having
to actually carry out the purification.
The above examples described where this technique has been applied
to elucidate the potential for purification of rFVIII and the correlation
between the quality of rFVIII as determined by ZAP and the yield
from a chromatography process. The above examples are intended to
illustrate the invention and it is thought variations will occur
to those skilled in the art. Accordingly, it is intended that the
scope of the invention should be limited only by the claims below.