J_ID: BTPR Customer A_ID: BTPR483 Ed. Ref. No.: BTPR10-0117.R1 Date: 18-August-10
1
2
3
4
5
6
7 AQ1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
Stage:
Page: 1
Impact of Antibody Aggregation on a Flowthrough Anion-exchange
Membrane Process
Ajay R. Lajmi and Samuel Nochumson
Pall Life Sciences, 8780 Ely Road, Pensacola, FL 32514
Aym Berges
Wyatt Technology Corporation, 6300 Hollister Avenue, Santa Barbara, CA 93117
DOI 10.1002/btpr.483
Published online 00 Month 2010 in Wiley Online Library (wileyonlinelibrary.com).
The impact of typical anion-exchange flowthrough conditions on the IgG mass loading of
an anion-exchange membrane scale-down unit was investigated. High performance sizeexclusion chromatography and multiangle laser light scattering results suggested the presence of a small fraction of IgG aggregates with average radius [100 nm under anionexchange flowthrough conditions. The small filtration area presented by the 0.35 mL membrane volume MustangV Q coin limited the membrane throughput due to fouling from the
aggregates at higher antibody loading. Data in this report indicated that a 0.2 lm hybrid
polyethersulfone and polyvinylidene fluoride membrane in-line prefilter with a minimum filtration area of 20 sq cm alleviated the Mustang Q coin fouling. The combined cake filtration
and intermediate blocking model was proposed as the most likely membrane pore blocking
mechanism. Increasing the filtration area in the in-line prefilter resulted in higher IgG mass
throughput. Thus, using an appropriately sized in-line prefilter could provide more robust
C 2010
antibody throughput performance on scale-down membrane anion-exchange units. V
American Institute of Chemical Engineers Biotechnol. Prog., 000: 000–000, 2010
Keywords: membrane chromatography, antibody purification, anion-exchange chromatography,
antibody aggregation, membrane fouling, light scattering
R
Introduction
Many antibody purification processes involve an anionexchange chromatography step in the flowthrough mode for
contaminant removal (Fahrner et al., 2001; Shukla et al.,
2007). Membrane anion-exchange chromatography has
recently emerged as a promising technology that could enable processing of large antibody batches (Low et al., 2007).
Reports on new salt tolerant membrane chemistries for contaminant clearance have led to sustained interest in this technology (Riordan et al., 2009).
Ion-exchange membranes have been investigated for various applications in the past. Anion-exchange membranes
were used in plasma fractionation (Gebauer et al., 1997). Efficient capture of different types of gene therapy virus vectors directly from clarified mammalian cell culture
supernatants and cell lysates were demonstrated using such
membranes (Lajmi et al., 2006). Investigation of binding
characteristics of Densonucleosis virus with anion exchange
membranes supports the more favorable dynamic binding
capacities of such large biomolecules compared to beaded
chromatography adsorbents (Specht et al., 2004). The large
scale capture of plasmid DNA from a clarified E. coli cell
lysate with a concomitant four log removal of endotoxins
using strong anion exchange membrane demonstrated not
only the high dynamic binding capacity of anion-exchange
Correspondence concerning this article should be addressed to A. R.
Lajmi at ajay_lajmi@pall.com.
membranes but also the efficiency of such membranes to
bind large anionic contaminants (Zhang et al., 2003). However, only recently have they been considered as potential
alternatives to packed-bed anion exchange chromatography
for disposability, fast flow rates that are independent of binding capacity and smaller footprint. A comprehensive review
of current applications of membranes in downstream purification of recombinant proteins and antibodies was recently
published (van Reis and Zydney, 2007).
It has been observed that the anion-exchange small-scale
units developed high operational pressure at much lower
antibody mass throughputs compared to that observed with
larger scale units (Zhou and Tressel, 2006). Although
attempts have been made to explain high pressures across
scale-down anion-exchange membrane chromatography units
in the flowthrough mode, the root cause was not clear (Zhou
et al., 2006).
Monoclonal antibody (mAb) purification process platforms
typically involve anion-exchange chromatography in the
flowthrough mode for polishing or contaminant removal
(Shukla et al., 2007). Since most host cell proteins are more
acidic than the therapeutic human or humanized mAb and
many mAbs have isoelectric points between 8 and 9, the
anion-exchange flowthrough step is usually operated at a
loading pH between 7 and 8. However, under those loading
pH conditions, the antibody product may tend to aggregate
as the loading pH approaches its isoelectric point.
The presence of protein aggregates in protein drugs has
been implicated in adverse immunological reactions (Braun
C 2010 American Institute of Chemical Engineers
V
ID: thambikkanue
I
Black Lining: [ON]
1
I
Time: 21:39
I
Path: N:/3b2/BTPR/Vol00000/100115/APPFile/JW-BTPR100115
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
J_ID: BTPR Customer A_ID: BTPR483 Ed. Ref. No.: BTPR10-0117.R1 Date: 18-August-10
2
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
Stage:
Page: 2
Biotechnol. Prog., 2010, Vol. 00, No. 00
et al., 1997). It has also been reported that the immunogenicity of aggregated recombinant human interferon alpha 2b
depends on the aggregate size (Hermeling et al., 2005).
Membranes that are typically used in antibody downstream
purification processes are sensitive to fouling due to the presence of aggregates. Thus, formation of antibody aggregates
during a process is not only a major clinical safety concern
but it also impacts the lifetime of membranes used in different unit operations during antibody downstream purification.
The aim of this work was to investigate the effect of IgG
flowthrough process conditions on anion-exchange membrane operating pressures during antibody loading on a
scale-down unit, namely, the Mustang Q coin. Commercially
available polyclonal IgG derived from human plasma was
used in this study as a surrogate for mAb at typical at pH
8.0 and conductivity of 15 mS cm1 (150 mM NaCl).
Although typical anion-exchange chromatography loading
conductivities are lower than 8 mS cm1, a higher conductivity was used here to prevent IgG from binding to the
membrane. Factors such as pH and effective membrane filtration (surface) area were examined to study their effects on
membrane fouling. Size-exclusion chromatography (SEC)
with multiangle laser light scattering (MALS) detection and
quasi-elastic light scattering (QELS) detection was used to
probe IgG aggregation. Scanning electron microscopy (SEM)
was used to show fouling of the anion-exchange coin. Correlation of antibody aggregation with membrane fouling was
confirmed by comparing the trends of membrane operating
pressures vs. antibody throughputs with different membrane
fouling models. Finally, the effectiveness of an in-line
0.2 lm filter upstream of the anion-exchange coin was investigated for providing high IgG flowthrough mass loading.
Thus, a better understanding of the impact of anion-exchange
flowthrough loading conditions on antibody aggregation
could help in developing more robust scale-down methods.
experiments were performed either with an in-line prefilter
such as the FluorodyneV EX EDF membrane (Mini KleenpakTM 20 capsule) with 20 sq cm filtration area connected
upstream of the coin or without such an in-line prefilter as
noted in each case. The AKTA system pressure sensor was
used to monitor the pressure therefore; all pressure readings
reported here and referred to as operational pressure were in
fact the total system pressure.
A 5% solution of polyclonal human IgG was diluted from
50 to 5 mg mL1 using 1 M Tris pH 8.0 and 5 M NaCl to a
final buffer concentration of 25 mM Tris pH 8.0 þ 150 mM
NaCl. The diluted IgG solution was prefiltered through a 0.2
lm 47 mm SuporV membrane disc. Although the hold time
after this filtration and before loading on the coin was not
recorded, care was taken to maintain this hold time to
15 min or less. The typical loading time was 1.5 h for the
anion-exchange coin with a 20 sq cm in-line filter, 8.5 h
with the 20 sq cm in-line filter and 14 h with the 200 sq cm
in-line filter. The anion-exchange membrane coin was preconditioned with 20 MV of 1 M NaOH followed by 20 MV
of 1 M NaCl and finally with the loading buffer (25 mM
Tris pH 8.0 þ 150 mM NaCl) until the pH and conductivity
of the effluent from the Q membrane coin reached baseline.
The IgG concentration was measured spectrophotometrically
from its absorbance at 280 nm using a conversion factor of
0.761 mg mL1.
The Fluorodyne EX EDF membranes used in this study
contained two membrane layers, an upstream 0.2 lm asymmetric PES followed by a 0.2 lm polyvinylidene fluoride
membrane. These membranes were used in the following
formats as in-line prefilters: Mini Kleenpak syringe filter (2.8
sq cm filtration area), Mini Kleenpak 20 capsule (20 sq cm
filtration area) and Mini Kleenpak 200 capsule (200 sq cm
filtration area).
R
R
Size-exclusion chromatography
Materials and Methods
A 5% solution of polyclonal human IgG purified from
human plasma was obtained from SeraCare Life Sciences
(Milford, MA) and was used as received. Mustang Q membrane coins (Pall Life Sciences, Port Washington, NY)
henceforth referred to as the anion-exchange membrane coin
or the Q membrane coin with 0.35 mL membrane volume
(MV) were assembled in its stainless steel housing. Tris(hydroxymethyl)aminomethane hydrochloride (Tris HCl) and
Tris(hydroxymethyl)aminomethane (Tris base) were obtained
from Research Organics (Cleveland, OH). Sodium chloride
and sodium hydroxide 50% (w/w) were purchased from
VWR International (West Chester, PA). All anion-exchange
IgG flowthrough experiments were performed on an AKTA
Explorer 100 (GE Healthcare, Uppsala, Sweden).
Membrane chromatography
The anion-exchange membrane coin was connected to an
AKTA Explorer with the flow restrictor disconnected. The Q
membrane coin consisted of a 16-layer stack of 13 mm diameter, 0.8 lm polyethersulfone (PES) membrane with quaternary ammonium chemistry and a filtration area (frontal
surface area) of 1.6 sq cm with a MV of 0.35 mL. The flow
rate used during IgG loading was 3.5 mL min1 that corresponds to 10 MV/min or 1313 L/sq m h unless otherwise
noted. All membrane anion-exchange IgG flow through
ID: thambikkanue
I
Black Lining: [ON]
I
Time: 21:39
I
High performance SEC was performed following the
method described by Ahrer et al., 2004. In this study, a TSK
G3000SW column (60 cm 7.5 mm i.d., Tosoh Biosep,
Montgomerryville) was used with a Shimadzu Prominence
20 AT HPLC instrument that was connected to a dual wavelength SPD-20A spectrophotometer. The SEC sample injection volume of a 5 g L1 IgG was typically 10 lL. An
isocratic gradient over 45 min was used with a mobile phase
containing 25 mM Tris pH 8.0 þ 150 mM NaCl at 0.6 mL
min1 flow rate. The SEC peaks corresponding to the high
molecular weight (HMW) IgG aggregates and the monomer
were integrated to calculate the percentage of IgG monomer
in the aggregated form.
SEC with MALS detection
SEC-MALS analysis was performed on a DAWN
HELEOS (Wyatt Technology Corporation, Santa Barbara,
CA) detector using an Agilent 1200 HPLC system with the
UV detector set at 214 nm. Additionally, optional QELS detector was used to determine the radius of particles at or
below 10 nm in radius. Protein concentration was also determined using an Optilab rEX differential refractive index detector (Wyatt Technology Corporation, Santa Barbara, CA).
Data processing for calculating the average molecular
weights and average radii of IgG species corresponding to
the SEC peaks in the chromatograms was performed using
Path: N:/3b2/BTPR/Vol00000/100115/APPFile/JW-BTPR100115
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
J_ID: BTPR Customer A_ID: BTPR483 Ed. Ref. No.: BTPR10-0117.R1 Date: 18-August-10
Stage:
Page: 3
Biotechnol. Prog., 2010, Vol. 00, No. 00
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
3
the ASTRA software version 5.3.2.20 (Wyatt Technology
Corporation).
Turbidimetry
Turbidity was measured using a Hach Laboratory Turbidimeter (model 2100N; Hach Company, Loveland, CO) in
the ratio on mode with StablCalV stabilized Formazin turbidity standards. The IgG samples were incubated at room temperature without stirring. In the turbidity study, the following
buffers with 25 mM concentration and 150 mM NaCl were
used at pH 5.2, 6.0, and 10.0, respectively: sodium acetate,
2-(N-Morpholino)ethanesulfonic acid (MES) and 4-(Cyclohexylamino)-1-butanesulfonic acid (CABS). At pH 7.4, 8.0
and 8.9, 25 mM Tris buffer containing 150 mM NaCl was
used, respectively.
R
Each data point was average of three readings and the error
bars refer to the standard deviation from the average.
Scanning electron microscopy
Different sample preparation methods were used based on
the required type of image. When a surface image was
desired, a small piece of the membrane was cut and sputtered coated with either a thin gold layer or a gold and palladium 60:40 mixture before scanning. For a cross-sectional
image, a small piece of the membrane was frozen under liquid nitrogen and carefully fractured to expose the cross-section. Subsequently, the section of the membrane was
sputtered coated with a thin layer of gold or a gold and palladium 60:40 mixture before scanning. The images were
obtained on an ISI DS-130 scanning electron microscope (Japan) with secondary electrons detector at a voltage of 19 kV.
Membrane fouling models
P
1
Kc J02
1
¼
lnð1 Kb tÞ
Kb
P0 ð1 Kb tÞ
(1)
where P and P0 are operational pressure at time t or at t ¼
0. J and J0 are permeate flux in L/m2 s. Kb and Kc are
complete blocking constant (s1) and cake filtration constant
(m1).
Combined Cake Filtration and Intermediate Blocking
Model.
P
K c J0
¼ expðKi J0 tÞ 1 þ
ðexpðKi J0 tÞ 1Þ
Ki
P0
(2)
where Ki is the intermediate blocking constant (m1).
Combined Complete Blocking and Standard Blocking
Model.
1
ð1 Kb tÞ 1 þ K2Ks Jb0 lnð1 Kb tÞ
2
(3)
where Ks is the standard blocking constant (m1).
ID: thambikkanue
Combined Intermediate Blocking and Standard Blocking
Model.
P
¼
P0
Combined
Model.
Cake
P
¼
P0
expðKi J0 tÞ
Ks
ðexpðKi J0 tÞ 1Þ
1 2K
i
Filtration
and
(4)
2
Standard
Blocking
!
Ks J0 t 2
1
þKc J02 t
2
(5)
Results and Discussion
Five combined fouling models generated by Bolton et al.
(2006) that were derived from Darcy’s Law under constant
flow conditions were tested against experimentally observed
fouling curves.
Combined Cake Filtration and Complete Blocking
Model.
P
¼
P0
Figure 1. Effect of pH and time at 218C on IgG aggregation in
different pH buffers containing 150 mM NaCl.
I
Black Lining: [ON]
I
Time: 21:39
I
Influence of pH and concentration on IgG aggregation
Ion-exchange chromatography is typically performed
between pH 5.0 and 9.0, while anion-exchange chromatography is commonly operated between 7.0 and 9.0. Figure 1
shows the effect of pH and incubation time in different buffers containing 150 mM NaCl on IgG aggregation in a 5 mg
mL1 IgG solution as measured by turbidimetry. The typical
conductivity in an anion-exchange flowthrough step is \8
mS cm1 that corresponds to 50 mM NaCl, (Fahrner
et al., 2001) however, in this study much higher NaCl concentration was used. This was necessary to prevent any binding of some of the more acidic polyclonal IgG species to the
anion-exchange membrane. As a result, quantitative IgG
recoveries were obtained in this study. Although higher loading conductivity limits impurity clearance such as the viral
clearance ability, (Strauss et al., 2009) the scope of this
study was to identify reasons for high pressure limitations on
scale-down membrane anion-exchange capsule under IgG
nonbinding conditions. Thus, limiting operating pressure was
used for comparing IgG throughput. The narrow increase in
IgG turbidity ranged from 2.0 to 7.0 Nephelometric Turbidity Unit (NTU) suggesting low levels of IgG aggregation.
Moreover, turbidity increased with increase in pH from 5.0
until it reached a maximum at pH 8.0 but subsequently
decreased from pH 8.0 to 10.0. The turbidity maxima were
observed between pH 6.5 and 8.0. Since the IgG was polyclonal with isoelectric points in the range between 6.8 and
8.2, the pH range at which the turbidity maxima were
observed correlated with the IgG isoelectric point range. The
turbidity trend suggested that insoluble IgG aggregate
Path: N:/3b2/BTPR/Vol00000/100115/APPFile/JW-BTPR100115
317
318
319
320
321
322
323
324
325
326
C
327
O
328
L
329
O
330
R
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
F1
355
356
357
358
359
360
361
362
363
364
365
366
AQ2
367
368
369
370
371
372
373
374
375
376
377
378
379
380
J_ID: BTPR Customer A_ID: BTPR483 Ed. Ref. No.: BTPR10-0117.R1 Date: 18-August-10
4
Stage:
Page: 4
Biotechnol. Prog., 2010, Vol. 00, No. 00
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
C
416
O
417
L
418
O
419
R
420
Figure 2. (a) SEC chromatogram of a 5 mg mL21 polyclonal hIgG in 25 mM Tris pH 8.0 and 0.150 M NaCl. (b) Effect of pH on IgG
421
aggregation.
422
Percentage of high molecular weight (HMW) IgG aggregates and IgG monomers, as measured by area under the SEC peak at 18 and 25 min reten423
tion time, respectively.
424
425
426
427
IgG mass throughput on anion-exchange membrane coin
formation might be more favorable as the solution pH
428
in flowthrough mode
approached the IgG isoelectric point.
429
Analytical SEC HPLC was used as a complimentary techAntibody loading capacity in adsorptive chromatography
430
nique for monitoring the influence of pH on IgG aggregais commonly defined in g L1 of adsorber (Knudsen et al.,
431 F2 tion. Figure 2a shows a typical SEC chromatogram of a 5 g
2001) even in reference to membrane chromatography. How432
L1 IgG in 25 mM Tris pH 8.0 þ 150 mM NaCl. The self
ever, in this work, as in Zhou et al., 2006, it was defined in
433
associated, HMW IgG aggregates eluted in the column void
g/sq m so that the data could be consistently applied to
434
volume at 18.4 min whereas the dimer eluted at 21.2 min
membrane fouling models. A typical operating pressure pro435
and the monomer at 25.4 min retention time. Integration of
file of a coin with 5 g L1 IgG in 25 mM Tris pH 8.0 þ
436
the SEC peaks corresponding to the HMW IgG aggregates
150 mM NaCl in the flowthrough mode is shown in Figure
437
and the monomer provided a quantitative measure IgG
3. Human polyclonal IgG consists of a diverse range of iso438
aggregates in solution at a given pH.
electric points from 4.7 to 9.0 (Szenczi et al., 2006). At an
439
As the pH was increased from 5.0 to 10.0, the HWM IgG
IgG mass loading of 1800 g/sq m and at a constant volumet440
aggregation increased from 8.7 to over 12.0% as shown in
ric flow rate of 3.5 mL min1 (1313 L/sq m h) the operating
441
Figure 2b. A concomitant decrease in IgG monomer was
pressure on the coin reached 2.1 105 Pa (0.2 MPa), an
442
observed suggesting that the IgG monomers were converted
upper limit in many antibody downstream purification manu443
to HMW aggregates as the pH increased from pH 5.0 to 9.0
facturing processes. Quantitative IgG recovery was observed
444
possibly through self-association.
in the membrane flowthrough pool based on the absorbance
ID: thambikkanue
I
Black Lining: [ON]
I
Time: 21:39
I
Path: N:/3b2/BTPR/Vol00000/100115/APPFile/JW-BTPR100115
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
F3
502
503
504
505
506
507
508
J_ID: BTPR Customer A_ID: BTPR483 Ed. Ref. No.: BTPR10-0117.R1 Date: 18-August-10
Stage:
Page: 5
Biotechnol. Prog., 2010, Vol. 00, No. 00
5
509
510
511
512
513
514
515
516
517
518
519
520
521
522
Figure 3. IgG mass throughput as a function of operating
523
pressure in the flowthrough mode on the Q mem524
brane coin with 5g L21 IgG in 25 mM Tris pH 8.0
1 150 mM NaCl at a constant flow rate of 3.5 mL
525
min21 or 1313 L/sq m h.
526
527
528
at 280 nm. On the contrary, an exponential increase in pres529
sure beyond an IgG mass loading of 1800 g/sq m suggested
530
fouling of the membrane pores by insoluble aggregates.
531
Since SEC analysis in Figure 2b indicated the possibility of
532
formation of larger, self-associated IgG molecules under the
533
loading pH and conductivity conditions, further investigation
534
of the size of those IgG molecules and its effects on the
535
membrane mass loading was pursued. The IgG mass
536
throughput of 1800 g/sq m corresponds to 720 g L1 of
537
membrane. To take advantage of the disposability of anion538
exchange membranes and improve the process economics a
539
much higher IgG throughput would be required (Zhou and
540
Tressel, 2006). One would require 13,500 g IgG/L of mem541
brane to process an entire batch of IgG from a 15,000 L cell
542
culture harvest at 1 g L1 IgG concentration after protein A
543
and cation exchange chromatography purification, assuming
544
90% overall yield (Zhou and Tressel, 2006).
545
546
IgG aggregate size and impact on filtration area
547
548
It is noteworthy that the IgG solution was filtered through
549
a 0.2 lm filter before loading on the anion-exchange mem550
brane coin as described in the methods section. It therefore
551
appears that the membrane fouling IgG aggregates were
552
formed during loading. Typically, membrane fouling on the
553
coin occurred beyond 30 min of loading at the flow rate
554
indicated in Figure 3. The time dependent turbidity profile at
555
different loading pH shown in Figure 1 indicated that the
556
turbidity maxima occurred beyond 30 min of room tempera557
ture incubation between pH 7.0 and 8.0. This correlated with
558
the onset of fouling shown in Figure 3.
559
Although using an in-line prefilter connected upstream of
560
the Q coin could improve the IgG mass throughput, it was
561
observed that a significant factor that limited the IgG mass
562
throughput under flowthrough anion-exchange conditions
563
was the filtration area of the in-line prefilter. As seen in Fig564
F4 ure 4a, the pressure profile with the 2.8 sq cm Fluorodyne
565
EX EDF prefilter followed an exponentially increasing trend
566
similar to that when the Q coin was used without an in-line
567
filter. Although the pressure profile shows slightly worse per568
formance with this in-line filter compared to the coin without
569
the filter, it can be concluded that within the variability of
570
the testing data this in-line filter did not provide much bene571
fit in improving the IgG throughput. As the in-line prefiltra572
tion area was increased to 20 and 200 sq cm, the IgG mass
ID: thambikkanue
I
Black Lining: [ON]
I
Time: 21:39
I
Figure 4. (a) Effect of filtration area of Fluorodyne EX EDF
in-line prefilter on Q membrane coin IgG throughput at a constant flow rate of 3.5 mL min21 or 1313
L/sq m h with respect to the coin. (b) Effect of filtration area of Fluorodyne EX EDF in-line prefilter
capacity at a constant flow rate of 3.5 mL min21 or
1313 L/sq m h with respect to the coin.
throughput increased to 20,000 g/sq m and [80,000 g/sq m
at 0.2 MPa respectively. A large 200 sq cm in-line prefilter
is oversized and impractical for use with the Q membrane
coin, the smaller, 20 sq cm in-line prefilter is a viable option
for improving the flowthrough IgG mass throughput on the
coin.
One reason for observing improved IgG throughput on the
anion-exchange coin with increased in-line filtration area
could be due to the in-line filtration capacity. As the filtration area increases it is conceivable that the filtration
capacity for entrapment of aggregates increases. This was
demonstrated in Figure 4b where at a total operational pressure of 0.2 MPa, the 2.8 sq cm in-line filter had a capacity
of 560 g/sq m. while the 20 sq cm filter had a capacity of
1,600 g/sq m. However, comparable IgG mass loading on
the 200 sq cm filter was not possible due to limitations on
availability of IgG.
Since the IgG filtration throughput data suggested the possibility of the IgG aggregates as the root cause of membrane
fouling, an investigation of the size of those aggregates was
pursued using SEC MALS analysis. This analysis shown in
Figure 5 revealed that the unfiltered IgG and the 0.2 lm filtered IgG chromatograms were almost identical except in the
aggregate region labeled as peak 5 corresponding to the
retention time between 11 and 12 minutes. The overlay-plot
Path: N:/3b2/BTPR/Vol00000/100115/APPFile/JW-BTPR100115
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
F5
634
635
636
J_ID: BTPR Customer A_ID: BTPR483 Ed. Ref. No.: BTPR10-0117.R1 Date: 18-August-10
6
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
C
662
O
663
L
664
O
R
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
Stage:
Page: 6
Biotechnol. Prog., 2010, Vol. 00, No. 00
Figure 5. Average IgG radius of monomer, dimer and aggregates using MALS and QELS detectors.
of average aggregate radius vs. retention time shown in Figure 5 shown on the secondary y-axis indicated that the average IgG aggregate radius was [100 nm as measured by the
MALS detector. Although a different 0.2 lm filter was used
here, Figure 5 demonstrated that the MALS detector could
measure reduction in IgG aggregates compared to the unfiltered IgG solution. Thus, irrespective of the type of 0.2 lm
filter a measurable reduction in IgG aggregation was
observed.
Evaluating the root mean square (RMS) radii of various
IgG species in solution indicated that in the unfiltered IgG
solution at pH 8.0 and 150 mM NaCl, the HMW IgG aggregate had a RMS radius of 0.314 lm (314 nm). This species
corresponded to the shoulder region in the SEC chromatogram in Figure 5 between 11 and 13 min retention time.
Such a large IgG aggregate would most likely block smaller
pores in the 0.8 lm nominal pore size anion exchange membrane thereby resulting in severe membrane fouling under
high IgG mass loading conditions. It is worthy to mention
that it is possible that some larger IgG aggregates were
trapped by the guard column during SEC-MALS analysis.
Dynamic light scattering in the batch mode could provide information on the IgG aggregate size distribution pattern,
however, that will be subject of another paper on a related
topic.
Interestingly, a 0.2 lm filtered IgG solution showed that
the RMS radius of the high molecular weight IgG aggregate
was only 58.6 nm indicating removal of the larger aggregates by the filter. Such a filtered IgG solution may alleviate
some of the severe membrane fouling issues. Although all
IgG solutions in this study were filtered through a 0.2 lm
PES filter before loading on an anion-exchange membrane
ID: thambikkanue
I
Black Lining: [ON]
I
Time: 21:40
I
coin, fouling occurred beyond 1800 g/m2 IgG mass
throughput. This suggested that some of the IgG aggregates
formed larger self-associated species over time even after
0.2 lm filtration as observed in Figure 1 where turbidity
increased over time at pH 8.0. The IgG hold time involving
filtration before loading on the anion-exchange membrane
coin was typically 15 min or less compared to the loading
time of 1.5 h for the coin without an in-line filter, 8.5 h for
the coin with the 20 sq cm filter and 14 h for the coin with
the 200 sq cm filter. Thus, the hold time involving filtration
of the IgG solution before loading on the membrane coin
with or without the in-line filter was insignificant compared
to the loading time. Even though the HMW IgG aggregates
comprised only 0.1% of all the IgG species in solution, they
likely blocked the smaller pores in the 0.1 lm range with a
gradual build-up of aggregates at those pores at higher IgG
mass loading. At high IgG mass loading, pore blocking of
such smaller pores could ultimately result in cake layer formation with a small effective filtration area of 1.6 sq cm on
a Q coin.
A more visual confirmation of membrane fouling due to
large IgG aggregates was evident from Figure 6a. A crosssectional view of the fouled first upstream layer of the 16layer coin membrane stack (Figure 6b) showed that IgG
fouling penetrated only the top 3 lm of that membrane
region. The other downstream membrane layers showed no
visible fouling by SEM (data not shown). Thus, Figures 6a,b
suggested that a likely membrane fouling mechanism could
involve initial pore blocking of smaller pores through sizeexclusion of the HMW aggregates in the upstream region of
the top membrane layer in the Q coin followed by cake
filtration.
Path: N:/3b2/BTPR/Vol00000/100115/APPFile/JW-BTPR100115
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
F6
755
756
757
758
759
760
761
762
763
764
J_ID: BTPR Customer A_ID: BTPR483 Ed. Ref. No.: BTPR10-0117.R1 Date: 18-August-10
Stage:
Page: 7
Biotechnol. Prog., 2010, Vol. 00, No. 00
7
aggregates accumulating above those that reside at the pore
Membrane fouling models
765
entrance. Cake filtration typically is a phenomenon where
766
Recently, membrane fouling models described in the exthe aggregates form a layer of increasing thickness thus
767
perimental methods section were also used to model fouling
forming a resistance to the flow. Standard blocking on the
768
of depth filters during filtration of harvested cell culture fluid
contrary pertains to entrapment of aggregates on the walls of
769
(Lutz et al., 2009). Five membrane pore blocking models
the pores.
770
were used in the present study to fit the experimental pres771 F7 sure vs. time curve as shown in Figure 7. It is important to
The best fit was determined by minimizing the sum of
772
squared residuals (SSR). The combined cake filtration and
note that these models were not used as predictive tools but
773
complete blocking model as well as the combined cake filrather to fit the observed experimental fouling curves. Com774
tration and intermediate blocking model gave nearly similar
plete blocking refers to aggregates residing at the pore en775
SSR values. However, the best fit was observed with the
trance while intermediate blocking can be viewed as
776
combined cake filtration and intermediate blocking model as
777
determined by the smallest SSR value (Table 1). This sug778
gested that the combined cake filtration and intermediate
779
blocking model was a more likely fouling mechanism than
780
the other models. The implication of this fouling mechanism
781
is that membrane fouling is most likely initiated as aggregate
782
deposition at the pore entrance, however, prolonged antibody
783
loading results in subsequent cake layer formation. A visual
784
manifestation of this fouling mechanism is evident from the
785
SEM images. Comparison of the fouled and unfouled mem786
brane top views in Figures 6a,c indicated widespread cake
787
layer formation while the membrane cross-sectional view in
788
Figure 6b showed aggregate deposition only 3 lm deep
789
suggesting intermediate blocking as well.
790
Contributions of the component models to the combined
791
models can be evaluated by comparison of the magnitudes
792
of the fouling constant values in Table 1. For example, in
793
the combined cake filtration and complete blocking model,
794
the cake filtration constant, Kc was three orders of magnitude
795
higher than the complete blocking constant, Kb indicating
796
that caking was a major component of the combined model.
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
Figure 7. Pressure vs. time experimental data fit for the typical membrane pore blocking models.
815
Model 1: Combined cake filtration and complete blocking.
816
Model 2: Combined cake filtration and intermediate blocking.
817
Figure 6. Scanning electron micrograph.
Model 3: Combined complete blocking and standard blocking.
818
Model 4: Combined intermediate and standard blocking. Model
(a) Top view of the first layer of the Q membrane coin fouled
5: Combined cake filtration and standard blocking. Data for 5 g
with IgG during processing without an in-line prefilter. (b)
819
L1 IgG operated at a constant flow rate of 3.5 mL min1 or
Cross-section viewed from the bottom. (c) Top view of an un820
1313 L m2 h on the Q membrane coin.
fouled membrane.
821
822
Table 1. Membrane Fouling Model Analysis and Parameters
823
824
Fouling Model
Model Fit Error, SSR
Fouling Constants
825
Combined cake filtration and complete blocking
2.06 101
Kb ¼ 1.00 103 s1, Kc ¼ 1.18 m1
826
Combined cake filtration and intermediate blocking
1.84 101
Ki ¼ 4.00 104 m1, Kc ¼ 9.11 104 m1
Combined complete blocking and standard blocking
1.98 102
Kb ¼ 1.00 105 s1, Ks ¼ 6.5 103 m1
827
2
3
Combined
intermediate
and
standard
blocking
1.04
10
K
m1, Ks ¼ 1.00 107 m1
i ¼ 1.75 10
828
2
4
1
4
1
1.45 10
Combined cake filtration and standard blocking
ID: thambikkanue
I
Black Lining: [ON]
I
Time: 21:40
I
Kc ¼ 9.00 10
m , Ks ¼ 8.00 10
Path: N:/3b2/BTPR/Vol00000/100115/APPFile/JW-BTPR100115
m
829
830
831
832
833
834
835
836
837
838
839
840
841
T1
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
C
874
O
875
L
876
O
R
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
J_ID: BTPR Customer A_ID: BTPR483 Ed. Ref. No.: BTPR10-0117.R1 Date: 18-August-10
8
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
Stage:
Page: 8
Biotechnol. Prog., 2010, Vol. 00, No. 00
It is noteworthy to mention that antibodies with different
isoelectric points and different loading pH and conductivities
may lead to other fouling patterns. For instance, more acidic
or basic antibodies compared to the polyclonal antibodies
used in this study may be involved in other intermolecular
associations under the given loading conditions. It is therefore possible that some of the other fouling models mentioned in this work may show a better fit with the empirical
fouling curves of such antibodies.
Conclusions
In this work, fouling of the scale down anion-exchange
membrane coin was studied under antibody flowthrough
anion-exchange chromatography conditions. Data presented
in this article indicates that under those conditions, since the
loading pH was in the proximity of the antibody isoelectric
point, the propensity towards formation of self-associated
IgG aggregates was greater.
Investigation of the average radius of IgG aggregates with
SEC-MALS suggested that the IgG aggregates were large
enough to cause pore plugging of some of the smaller pores in
the anion-exchange membrane coin resulting in membrane
fouling at high antibody mass loading. The SEM images of the
fouled top layer of the anion-exchange membrane coin showed
that pore blocking occurred only a few microns deep in the
top membrane layer. A 0.2 lm in-line prefilter of 20 sq cm filtration area or higher was found to alleviate fouling on the
anion-exchange membrane coin. The use of 0.2 lm in-line filters for chromatography column protection is common practice
in downstream purification manufacturing suites. Thus, using
them at bench scale would result in more robust scale down
methods for antibody polishing applications with anionexchange membrane adsorbers. To the best of our knowledge,
this is the first reported investigation of the effect of anionexchange flowthrough chromatography conditions on the membrane adsorber antibody mass throughput and correlation with
antibody aggregation using MALS detection.
Acknowledgments
The authors acknowledge the technical assistance from Shiying Zhang, Alla Krivosheyeva, and Anil Kumar. They thank
Donald Simonton and Edward Magowan for scanning electron
microscope imaging. Lastly, they are grateful for the support of
this work from Ralf Kuriyel, Steven Pearl, and Richard Gutman. Mustang, Supor and Acrodisc are registered trademarks
of Pall Corporation. AKTA is a registered trademark of GE
Healthcare. DAWN, Optilab and ASTRA are registered trademarks of Wyatt Technology Corporation. StableCal is a registered trademark of Hach Company.
Literature Cited
Ahrer K, Buchacher A, Iberer G, Jungbauer A. Detection of aggregate formation during production of human immunoglobulin G by
means of light scattering. J Chromatogr A. 2004;1043:41–46.
ID: thambikkanue
I
Black Lining: [ON]
I
Time: 21:40
I
Bolton G, LaCasse D, Kuriyel R. Combined models of membrane
fouling: development and application to microfiltration and ultrafiltration of biological fluids. J Membr Sci. 2006;277:75–84.
Braun A, Kwee MA, Labow J, Alsenz J. Protein aggregates seem to
play a key role among the parameters influencing the antigenicity
of interferon alpha in normal and transgenic mice. Pharm Res.
1997;14:1472–1478.
Fahrner RL, Knudsen HL, Basey CD, Galan W, Feuerhelm D, Vanderlaan M, Blank GS. Industrial purification of pharmaceutical
antibodies: development, operation and validation of chromatography processes. Biotechnol Gen Eng Rev. 200l;18:301–327.
Gebauer KH, Thommes J, Kula MR. Plasma protein fractionation
with advanced membrane adsorbents. Biotechnol Bioeng. 1997;
54:181–189.
Hermeling S, Aranha L, Damen JM, Slijper M, Schellekens H,
Crommelin DJ, Jiskoot W. Structural characterization and immunogenicity in wild-type and immune tolerant mice of degraded
recombinant human interferon alpha 2b. Pharm Res. 2005;22:
1997–2006.
Knudsen HL, Fahrner RL, Xu Y, Norling LA, Blank GS. Membrane
ion-exchange chromatography for process-scale antibody purification. J Chromatogr A. 2001;907:145–154.
Lajmi AR, Kutner R, Reiser J. A membrane chromatography application: a rapid, high capacity gene therapy vector purification
tool. In: Shukla AA, Etzel M, Gadam S, editors. Process Scale
Bioseparations for the Biopharmaceutical Industry. Boca Raton:
Taylor Francis; 2006:541–564.
Low D, O’Leary R, Pujar NS. Future of antibody purification. J
Chromatogr B. 2007;848:48–63.
Lutz H, Abbott I, Blanchard M, Parampalli A, Setiabudi G, Chiruvolu V, Noguchi M. Considerations for scaling-up depth filtration
of harvested cell culture fluid. Biopharm Int. 2009;22:58–66.
Palacio L, Ho CC, Zydney AL. Application of a pore-blockage
cake-filtration model to protein fouling during microfiltration.
Biotechnol Bioeng. 2002;79:260–270.
Riordan W, Heilmann S, Brorson K, Kannan S, He Y, Etzel M.
Design of salt tolerant membrane adsorbers for viral clearance.
Biotechnol Bioeng. 2009;103:920–929.
Szenczi A, Kardos J, Medgyesi GA, Zavodszky P. The effect of solvent environment on the conformation stability of human polyclonal IgG in solution. Biologicals. 2006;34:5–14.
Shukla AA, Hubbard B, Tressel T, Guhan S, Low D. Downstream
processing of monoclonal antibodies—application of platform
approaches. J Chromatogr B. 2007;848:28–39.
Specht R, Han B, Wickramasinghe SR, Carlson JO, Czermak P,
Wolf A, Reif OW. Densonucleosis virus. Biotechnol Bioeng.
2004;88:465–473.
Strauss DM, Gorrell J, Plancarte M, Blank GS, Chen Q, Yang B.
Anion exchange chromatography provides a robust, predictable
process to ensure viral safety of biotechnology products. Biotechnol Bioeng. 2009;102:168–175.
van Reis R, Zydney A. Bioprocess membrane technology. J Membr
Sci. 2007;297:16–50.
Zhang S, Krivosheyeva A, Nochumson S. Large-scale capture and
partial purification of plasmid DNA using anion-exchange membrane capsules. Biotechnol Appl Biochem. 2003;37:245–249.
Zhou JX, Tressel T. Basic concepts in Q membrane chromatography
for large scale antibody production. Biotechnol Prog.
2006;22:341–349.
Zhou JX, Tressel T, Gottschalk U, Solamo F, Pastor A, Dermawan
S, Hong T, Reif O, Mora J, Hutchison F, Murphy M. New Q
membrane scale-down model for process-scale antibody purification. J Chromatogr A. 2006;1134:66–73.
Manuscript received Mar. 26, 2010. and revision received Jul. 21, 2010.
Path: N:/3b2/BTPR/Vol00000/100115/APPFile/JW-BTPR100115
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
AQ3
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020