Monitoring Product Attributes, Media Components,

and Metabolites with On-Line Liquid Chromatography

Rick Cooley, Dionex Corporation, Sunnyvale, CA, USA

Introduction

Experimental

Beginning in 2001, the US FDA took actions to encourage changes

in pharmaceutical manufacturing that would lead to the incorporation
of methods of continuous improvement and quality control that have
been used in other high technology industries. These methods involve
moving from a mode of testing-in quality to the more desirable method
of quality by design. To stimulate these changes, the US FDA issued two
guidance documents: Pharmaceutical cGMPs for the 21st Century;
A Risk Based Approach and PATA Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance. A desired
goal of the PAT framework is to design and develop well-understood
processes that will consistently ensure a predefined quality at the end of
the manufacturing process.

Direct detection of amino acids and carbohydrates in the on-line LC

analyzer is possible using integrated pulsed amperometric detection and
an LC compatible with up to 500 mM NaOH. In this mode of detection,
the applied potential is rapidly stepped through a series of potentials
designed to maximize analyte detection while minimizing deleterious
electrode effects. A chromatogram of common amino acids and glucose
in a diluted cell culture media is shown in Figure 1.

Pharmaceutical manufacturing is increasingly using biotechnologybased processes for formation of product molecules. Monitoring and
controlling these processes can be extremely challenging due to the
complexity of the cellular mechanisms and pathways that govern cell
growth and product formation. Individual amino acids, carbohydrates,
vitamins, trace metals, and metabolites can have a significant effect
on product quality and yield even when present in low levels.1 Quantification of these individual components can lead to a much greater
understanding of the cell culture process and the underlying cellular
mechanism that is leading to changes in the product. Identification of
these critical individual components and utilization of on-line analytical
methods with sufficient specificity and sensitivity to quantitate them can
lead to feedback control mechanisms that can continuously drive the
cell culture to produce the desired product quality and maximize yield.
This increased level of detailed understanding can be extremely valuable
in the process development stage of a pharmaceutical product when the
optimum process conditions are being identified and fine tuned, as well
as in production when troubleshooting of an unexpected event may
be required.
The complexity of a typical bioreactor, where thousands of molecules
are present2 during normal operation, creates a situation where measurement specificity and low-level detection limits are extremely important. On-line liquid chromatography (LC) is an analytical technique that
is uniquely suited to the challenges of bioreactor monitoring. LC with
integrated pulsed amperometric detection has the capability of directly
analyzing these complex mixtures for many of these important components, such as amino acids and carbohydrates in cell culture media.3

AminoPac PA10 Analytical, 2 250 mm

AminoPac PA10 Guard, 2 50 mm
Eluent:
NaOH/Acetate gradient
(see Application Update 152 for
gradient program)5
Flow Rate: 0.25 mL/min
Temperature: 30 C
Inj. Volume: 25 L
Detection:
IPAD
Columns:

400

12
13
17

11

nC
1
5

23
70

Peaks:

10

10
9

15
14 16

20

25

22

30

35

Minutes

14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.

Isoleucine
Leucine
Methionine
Histidine
Phenylalanine
System
Glutamate
Aspartate
Cystine
Tyrosine
Tryptophan

23

20 21
19

6
15

18

1. Arginine
2. Ornithine
3. Methionine
sulfoxide
4. Glucose
5. Glutamine
6. Asparagine
7. Alanine
8. Threonine
9. Glycine
10. Valine
11. Hydroxyproline
12. Serine
13. Proline

24

40

45

50

55

60

65
22981-01

Figure 1. Typical chromatogram of common amino acids and glucose in cell

culture media. Sample diluted tenfold.

The ability of the on-line LC to selectively monitor individual amino

acids is shown in Figure 2, where the asparagine concentration is
plotted versus time. Superimposed on this graph are the values obtained
from the laboratory-based amino acid analysis. The on-line assay
results compared quite favorably to the off-line method of analysis. Each
data point from the on-line analyzer is obtained approximately once per
hour. In contrast, the off-line amino acid assays were performed once
every 24 hours. It is important to note how clearly the change in asparagine concentration can be seen in the on-line data, whereas it is much
less obvious in the off-line analysis data.

Asparagine Concentration (mg/L)

11
Typical cell culture

10

Glucose controlled culture

9
8
7
6
5
4
3
2
1
0

24

48

72

96

120

144

168

192

216

IPAD on-line HPLC

AminoQuant off-line HPLC

250
200
150

24

48

72

96

120

144

168

192

Cutlure Time (hours)

26717

Figure 2. On-line asparagine concentration versus time compared to off-line

asparagine concentration values.

288

312

Figure 3. Comparison of on-line glucose feedback control with off-line

manual control.

50

264

26718

100

240

Culture Time (hours)

The ability to control the glucose concentration with the on-line LC

analyzer was used to study the impact of glucose level on alanine
production. Increased levels of alanine is an indication of overflow metabolism of glucose. Prevention of this overflow metabolism should be
readily apparent by the continuous monitoring of alanine concentration
while performing feedback control of the glucose concentration using
the on-line LC. The results are shown in Figure 4.

350
300

In another experiment, the ability of the on-line LC to perform feedback

control of the glucose concentration in the bioreactor was tested. To
do this, the capability of the analyzers software to generate alarms was
used. An alarm was configured in the software to be triggered whenever
the glucose concentration dropped below the alarm setpoint value. This
low alarm was set to close a relay in the analyzers controller whenever the alarm was active. This relay provided power to a peristaltic
pump which would feed sterilized glucose solution to the bioreactor. A
comparison of the bioreactor glucose concentration control with the online LC to the conventional control method that involved grab samples,
off-line analysis, and manual addition of glucose is shown in Figure 3.
The ability to achieve much improved glucose control, even with simple
on-off control, is apparent in the automatic mode.

Glucose Concentration (g/L)

The process development area of Genentech in San Francisco, CA,

USA used on-line LC with integrated pulsed amperometric detection to
monitor the concentration of amino acids and glucose in their development bioreactors, developing an understanding of parameters critical
to cellular metabolism across several cell lines.4 The on-line LC was
connected to two separate bioreactors through the analyzers optional
sample stream selection interface (SS80). This device uses a Rheodyne
valve to multiplex up to seven separate sample streams per valve.
The SS80 can accommodate up to three of the multiport valves, thus
providing the means to multiplex up to 21 separate sample streams per
on-line chromatograph. The sterility of the bioreactor was maintained
using either a sample probe inside the bioreactor that uses a 0.2 m
ceramic filter or a tangential flow filter located outside the bioreactor that
contains a sterilizing disk filter. Sample flow from the bioreactor to the
on-line LC analyzer was provided by a peristaltic pump. Since the analyzer is not sterile downstream of the sterile barrier, the sample lines are
flushed automatically with caustic solution to minimize bacterial growth
within the sample lines that could potentially lead to plugging. The online LC analyzer was able to perform the analysis of 19 amino acids and
glucose with an overall analysis time of approximately 65 min, including
regeneration of the column between samples.

Monitoring Product Attributes, Media Components, and Metabolites with On-Line Liquid Chromatography

Acknowledgement

1800

The author gratefully acknowledges the data provided for this presentation by Ms. Tina Larson of the Genentech Process Development group
in South San Francisco, CA, USA.

Typical culture (off-line data)

Alanine Concentration (mg/L)

1500

Glucose controlled, 500 mg/L

Glucose controlled, 50 mg/L

1200

References

900
600
300
0

24

48

72

96

120

144

168

192

Culture Time (hrs)

26719

Figure 4. On-line alanine concentration versus time with feedback control of

glucose concentrations at 50 mg/L and 500 mg/L.

Conclusions
Integrated pulsed amperometric detection provides an excellent detection method for quantification of amino acids and carbohydrates in
bioreactors without pre- or post-column derivatization. This detection
method was used in an on-line LC to provide automated analysis of
19 amino acids and glucose in a mammalian cell culture bioreactor
approximately once per hour. The on-line amino acid values compared favorably with off-line analysis methods except for high levels
of glutamine, which occurred at the beginning of a bioreactor run, and
alanine, which was not well resolved from glucose at high glucose
concentrations. The additional data generated by the on-line LC
provided increased understanding of metabolism differences across cell
lines with far less labor than would have been required using off-line
analysis. The increased amount of data allowed derived values, such
as amino acid uptake rate, to be predicted with increased accuracy. The
higher data frequency also made transient events that occurred in the
bioreactor much more readily apparent.

1) Jerums, M.; Xiaoming, Y. BioProcess International Supplement, June

2005, 3844.
2) Galliher, P. The Potential Benefits of PAT for Biologics Manufacturing,
Presentation I-124, IFPAC Conference, Arlington, VA, January 11,
2005.
3) Jandik, P.; Clarke, A. P.; Avdalovic, N.; Anderson, D. C.; Cacia, J.
Analyzing Mixtures of Amino Acids and Carbohydrates Using BiModal Integrated Amperometric Detection. J Chromatogr., B 1999
732, 193201.
4) Larson, T. M.; Gawlitzek, M.; Evans, H.; Albers, U.; Cacia, J.
Biotechnol. and Bioeng. 2002, 77 (5), 553563.
5) Dionex Corporation. An Improved Gradient Method for the
AAA-Direct Separation of Amino Acids and Carbohydrates in
Complex Sample Matrices; Application Update 152, LPN 1768.
Sunnyvale, CA 2006.

AAA-Direct is a trademark and AminoPac is a registered trademark of Dionex Corporation

North
America
Dionex Corporation

Europe

Asia Pacific

1228 Titan Way

P.O. Box 3603
Sunnyvale, CA

94088-3603

(408) 737-0700

Austria (43) 1 616 51 25 Benelux (31) 20 683 9768; (32) 3 353 4294
Denmark (45) 36 36 90 90 France (33) 1 39 30 01 10 Germany (49) 6126 991 0
Ireland (353) 1 644 0064 Italy (39) 02 51 62 1267 Switzerland (41) 62 205 9966
United Kingdom (44) 1276 691722

Australia (61) 2 9420 5233 China (852) 2428 3282 India (91) 22 2764 2735
Japan (81) 6 6885 1213 Korea (82) 2 2653 2580 Singapore (65) 6289 1190
Taiwan (886) 2 8751 6655

U.S./ Canada (847) 295-7500

South America

Brazil (55) 11 3731 5140

www.dionex.com

LPN 2376-01 10/09

2009 Dionex Corporation