Protein Purification Process Engineering

BIOPROCESS TECHNOLOGY
Series Editor
W. Courtney McGregor
Xoma Corporation
Berkeley, California
1. Membrane Separations in Biotechnology, edited by W. Courtney

PROTEIN
McGregor
2. Commercial Production of Monoclonal Antibodies: A Guide for Scale-
Up, edited by Sally S. Seaver
3. Handbook on Anaerobic Fermentations, edited by Larry E. Erickson
PURIFICATION
and Daniel Yee-Chak Fung
4. Fermentation Process Development of Industrial Organisms, edited by
Justin O. Neway
5. Yeast: Biotechnology and Biocatalysis, edited by Hubert Verachtert
PROCESS
and Rene De Mot
6. Sensors in Bioprocess Control, edited by John V. Twork and Alexan-
der M. Yacynych
ENGINEERING
7. Fundamentals of Protein Biotechnology, edited by Stanley Stein
8. Yeast Strain Selection, edited by Chandra J. Panchal
9. Separation Processes in Biotechnology, edited by Juan A. Asenjo
10. Large-Scale Mammalian Cell Culture Technology, edited by Anthony
S. Lubiniecki
11. Extractive Bioconversions, edited by Bo Mattiasson and Olle Holst
edited by
12. Purification and Analysis of Recombinant Proteins, edited by Ramnath Roger G. Harrison
Seetharam and Satish K. Sharma
University of Oklahoma
13. Drug Biotechnology Regulation: Scientific Basis and Practices, edited
by Yuan-yuan H. Chi& and John L. Gueriguian Norman, Oklahoma
14. Protein Immobilization: Fundamentals and Applications, edited by
Richard F. Taylor
15. Biosensor Principles and Applications, edited by Loic J. Blum and
Pierre R. Coulet
16. Industrial Application of Immobilized Biocatalysts, edited by Atsuo
Tanaka, Tetsuya Tosa, and Takeshi Kobayashi
17. Insect Cell Culture Engineering, edited by Mattheus F. A. Goosen,
Andrew J. Daugulis, and Peter Faulkner
18. Protein Purification Process Engineering, edited by Roger G. Harrison
ADDITIONAL VOLUMES IN PREPARATION
Recombinant Microbes for Industrial and Agricultural Applications,

edited by Yoshikatsu Murooka and Tadayuki Imanaka Marcel Dekker, Inc. New York* Basel* Hong Kong
208 Rothstein
ACS Symposium Series 427 (M. R. Ladisch, R. C. Willson, C. C.,

Painton, and S. E. Builder, eds.), American Chemical Society,
Washington, D.C., p. 66.
2. Clark, K. M., and Glatz, C. E. (1990). Protein fractionation by
precipitation with carboxymethyl cellulose, Downstream Processing
and Bioseparation, ACS Symposium Series 419 (J. P. Hamel, J. B.
Hunter, and S. K. Sikdar, eds.), American Chemical Society,
Washington, D.C., p. 170.
3. Parker, D. E., Glatz, C. E., Ford, C. F., Gendel, S. M., Suominen,
I., and Rougvie, M. A. (1990). Recovery of a charged-fusion protein Conventional Chromatography
from cell extracts by polyelectrolyte precipitation, Biotechnol.
Bioeng., 36: 467.
4. Zhao, J., Ford, C. F., Glatz, C. E., Rougvie, M. A., and Gendel, S.
M. (1990). Polyelectrolyte precipitation of B-galactosidase fusions John M. Simpson
containing poly-aspartic acid tails, J. Biotechnol., 14: 273. Lederle Laboratories, American Cyanamid Company, Pearl River, New York
5. Ford, C. f., Suominen, H., and Glatz, C. E. (1991). Fusion tails for
the recovery and purification of recombinant proteins, Prot. Expr.
Purif., , 2: 95.
I. Precipitation of a Soluble Affinity Complex
1. Bradshaw, A. P., and Sturgeon, R. J. (1990). The synthesis of I. INTRODUCTION
soluble polymer-ligand complexes for affinity precipitation studies,
The vast majority of protein purification processes involve at least one
Biotechnol. Tech., 4: 67.
chromatography step, and with good reason, for chromatography is very well
2. Chen, J. P. (1990). Novel affinity-based processes for protein
suited to the gentle separation of proteins of very similar structure. Indeed,
purification, J. Ferment. Koeng., 70: 199.
chromatography is often the key to the success of a separation process.
3. Fujii, M., and Taniguchi, M. (1991). Application of reversibly
Given the technical importance of chromatography and its relatively high
soluble polymers in bioprocessing, Trends Biotechnol., 9: 191.
expense (1) in terms of equipment and manpower, it is obvious that great
J. Metal Ions care should be taken to be sure that separations developed in the lab can be
1. Glusker, J. P. (1991). Structural aspects of metal liganding to scaled up to become commercial successes.
functional groups in proteins,Adv. Protein Chem., 42: 1. This chapter is intended to aid in the design and operation of large-scale
K. Affinity Precipitation conventional chromatography for the purification of proteins and peptides.
1. Morris, J. E., and Fisher, R. R. (1990). Complications encountered Conventional chromatography is taken to mean those operations that contact
using Cibacron Blue F3G-A as a ligand for affinity precipitation of a liquid mixture of proteins or peptides with a particulate adsorbent packed
lactic dehydrogenase, Biotechnol. Bioeng., 36: 737. in a bed for the purpose of separating a desired product from the mixture.
L. Precipitation Further characteristics of conventional chromatography are moderate
1. Glatz, C. E. (1990). Precipitation, Separation Processes in Biotech - pressure drop across the packed bed (<50 psig), batchwise operation, and a
nology (J. A. Asenjo, ed.), Marcel Dekker, New York, p. 329. cyclical sequence of processing steps. Specifically excluded from consider-
2. Richardson, P., Hoare, M., and Dunnill, P. (1990). A new biochemi - ation in this chapter are batch adsorption from a slurry, continuous
cal engineering approach to the fractional precipitation of proteins, chromatography, radial chromatography, HPLC, metal affinity chromatogra-
Biotechnol. Bioeng., 36: 354. Phy, dye ligand chromatography, and the use of adsorbents based on
3. Iyer, H. V., and Przybycien, T. M. (1993). Protein precipitation : antibodies or other biological interactions (affinity chromatography).
Effects of mixing on protein solubility, AIChE J., in press. Many types of chromatography are available to the process engineer.
The particular type chosen will depend on the type of protein being
Processed, the nature of the impurities, the yearly production required, the
209
Simp son Conventional Chromatography 211
210
position of the step in the process, and many other factors. As a guide to Generally the gels are coated with octyl or phenyl groups. The protein
making the proper choice, a brief description is given for each major type should be loaded in high salt concentration to encourage binding (11). For
available, not with the intent of being exhaustive but rather to draw attention this reason, the step usually follows ion exchange or ammonium sulfate
to those that are most useful and frequently applied. precipitation. Resolution in general is not as good as ion exchange, but the
method does hold the considerable advantage of being based on an entirely
different type of interaction compared to ion exchange. Bound protein is
II. TYPES OF CHROMATOGRAPHY eluted by lowering the salt concentration of the eluent or by raising the
A. Ion Exchange organic solvent content. Protein denaturation through too strong an
Ion exchange is one of the most frequently applied chromatographic steps. interaction with the matrix is sometimes a problem (12). In a few cases the
The reasons for this are not hard to understand. Ion exchange resin has a protein is eluted by water alone. The gel is more expensive than ion
high capacity to adsorb protein. Higher-molecular-weight proteins are exchange resin but less expensive than affinity or HPLC matrices. Capacity
adsorbed less on a weight basis than lower-molecular-weight proteins, but for is somewhat lower than ion exchange (13).
the average protein of 20,000 daltons a capacity of about 100 g/L of gel
should be expected (2). Since virtually all proteins are charged at one pH D. Chromatofocusing
or another, this method has very wide applicability. Even proteins with Chromatofocusing is a chromatographic technique, developed by Sluyterman
identical isoelectric points may be separated, for the protein attachment site (14) and commercialized principally by Pharmacia, that relies on a special ion
to the matrix is determined by the surface charge of the protein and the exchange resin (polyethyleneimine-agarose), similar to a weak base anion
proximity of other charges, not just by the integrated charge of the protein exchange gel, but with a continuous range of titratable groups from high to
(3,4). Ion exchange has good selectivity. Very often, conditions can be l ow pH. The column is equilibrated to a pH higher than the highest
found such that the desired protein is adsorbed while most impurities are expected isoelectric point (pl) of any of the sample components. A
not. By a judicious choice of the elution conditions the remaining impurities measured sample is injected onto the column, and a buffer of specially
may be separated from the protein of interest. The proper choice of gel, pH, formulated ampholytes of high buffering strength and low ionic strength is
ionic strength, and other factors may be guided by the text of Scopes (2) and fed to the column. The pH of the buffer must be less than the lowest pI
the electrophoretic titration method developed by Richey (5). Passage expected in the sample. The action of the ampholyte mixture on the gel
through an anionic gel is an extremely effective method to remove pyrogens creates a linear pH gradient in the gel bed. Proteins are eluted from the
from proteins (6). Depyrogenation is a must for all proteins derived from column at or slightly above their isoelectric point. Sample volume is limiting
recombinant E. coli (7). and the height of the bed is important (13). Separation of isoforms of the
same protein is possible with this technique, indicative of excellent resolving
B. Gel Permeation power. Resolution is affected most by gradient volume (15) and the range
Gel permeation chromatography separates proteins and peptides on the basis of pI to be covered. The major drawbacks are the high expense of the gel
of their molecular size (8). Separations can be achieved over the range of and special eluent, and the nuisance of having to remove the ampholyte after
102 to 10 7 daltons (9). It does have a place in larger-scale operations, but the chromatography is over. Application is usually limited to the final stages
only under restricted conditions. The major factors limiting its use are low of purification of a high-value-added product (16).
capacity and slow flow rate. If the impurities differ in molecular weight by
a factor of five or more, ultrafiltration should be examined as a low-cost, E. Affinity and Large-Scale HPLC
high-throughput alternative or at least pretreatment. If the product Affinity chromatography and process scale HPLC are specialized types of
commands a high price and must be ultrapure, a gel permeation step may be chromatography that tend to be very expensive but can be appropriate in some
advantageous close to the end of the process. Low annual requirements applications. A few comments will be made on the application of these tech-
(<10 kg/year) also favor GPC. Kelley (10) has estimated the cost of a nologies compared to conventional chromatography. For small-volume high-
typical GPC step as $5/g in 1987. value products containing closely related impurities, large-scale HPLC should
be considered if resolution by conventional chromatography is inadequate (17).
C. Hydrophobic Chromatography Scale-up from analytical separation to preparative scale to large scale is well
Hydrophobic chromatography is based on the interaction of hydrophobic established and has the advantage of minimizing development work and rush-
regions of the desired protein with a gel matrix possessing similar properties. ing the product to market. There is, however, an upper limit of about 2 to 10
kg/day productivity for each production machine, each of which costs hun-
212 Simpson Conventional Chromatography 213
dreds of thousands of dollars. To obtain the maximum resolution, loading separates the feed into a number of bands that emerge from the column.
must be carefully controlled, generally resulting in a large volume of dilute The weakly bound components emerge first and the strongly bound ones last.
eluent, which requires concentrating before further processing. The cost of This type of chromatography is typical of analytical, isocratic HPLC analysis.
the concentrating equipment is substantial. Expensive explosion-proof It calls for a very delicate balance of solvent-feed-matrix interactions that
equipment is required to safely handle the solvents present in the elution cannot be obtained easily in the case of proteins. Advantages are constant
buffers. The high pressures involved and the special requirements of the eluent composition, ease of operation, the possibility of repetitive injections
absorbent (low-micron-size particles, special coating, and particle-size without intervening reequilibrations, and good correspondence to design
uniformity) mean high operation and maintenance costs. Adsorbent lifetime models. Its disadvantages are limited use of the adsorbent capacity (it works
is a major concern. best at very light loading), dilution of the feed stream, the need for long
Affinity techniques using antibodies are attractive on the basis of their columns with accompanying high pressure drops and packing problems, the
exquisite resolution, which offers the possibility, on paper at least, of tendency to drift in repeatability due to the effects of irreversible adsorption
shortening the process considerably. The method selects for molecules with on the matrix, and the difficulty, if not impossibility in practical situations,
the proper folding to interact with antibodies attached to the gel. Counter- of finding the right combination of operating conditions for isocratic elution.
balancing these advantages are a number of serious disadvantages that tend Elution with an eluent of increasing strength (gradient elution) is a way
to limit industrial applications (18,19). The capacity of antibody columns is around the last disadvantage, but it introduces a new debit in the form of
low, typically around 2 g/L of gel (2), while the cost to prepare the antibodies added time for equilibration.
and attach them to the matrix is high. Robustness and column hygiene are
other concerns in a production setting. These columns represent a consider- B. On-Off Chromatography
able investment, which can be easily destroyed by mistakenly washing the On-off chromatography accounts for almost all of protein and large-scale
column with a strong reagent; they can also be irreparably fouled by peptide chromatography. The feed stream is loaded onto the column at
contacting with process streams that are too crude (20). If a series of related conditions that cause the desired component to stick strongly to the matrix.
proteins are to be developed into products, a new affinity column will be After all of the feed has been adsorbed, the composition of the eluent is
needed for each one unless the antibody interacts with a region common to changed so that the desired protein moves off the adsorbent and into the
all the members of the family. Convential chromatography is much easier fluid or mobile phase. The change in eluent composition depends on the
to adapt to a series of related new products. The final disadvantage of type of chromatography. For ion exchange, increasing ionic strength or
affinity techniques is the need to monitor leaching of the ligand from the changing pH is appropriate. In practice, changing pH is not very successful
column. Assays must be developed to follow the fate of the released ligand in supplying the required resolution unless the protein is adsorbed very
through the remainder of the process (16,20). The best position for affinity strongly in the first place (2). A much surer way to elute adsorbed proteins
chromatography is in the middle of the process. This placement maximizes selectively is to raise the ionic strength either stepwise or in a gradient. As
the resolving power of the method while protecting the adsorbent from the ionic strength increases, an ion in the eluent can either displace the
premature fouling. Any ligand that leaches off the column can be handled protein by directly occupying the site occupied by the protein or, if the ion
by the remaining refining steps. is a substrate for adsorbed enzyme, it can bind directly to the protein and
reduce the affinity of the protein for the adsorbent (2). This latter type of
III. MODES OF OPERATION elution can be quite specific and is called "affinity elution." In hydrophobic
chromatography (also called hydrophobic interaction chromatography), a
In addition to the type of chromatography, the mode of operation must also
method of elution that generally does not denature proteins is to lower the
be considered. i onic strength. This shifts the equilibrium of the protein between the
A. Elution Chromatography adsorbent and fluid phases so that at a certain ionic strength, the protein is
The elution mode is the one traditionally associated with chromatography. entirely in the fluid phase.
A small portion of material is introduced onto the column and adsorbs Chromatography of proteins by ion exchange with an increasing salt
lightly to it. As the eluting solvent flows through the column, the differing gradient is a typical application of this mode. Advantages are high use of gel
components in the feed desorb from the matrix, flow for a short time with capacity, ability to load dilute feed streams with no fear of heavy losses,
the solvent, and adsorb again. Each of the components in the feed differs concentrated product streams, greater flexibility in design of operating
in the strength of adsorption. Consequently, migration through the column conditions to optimize the separation, only short columns required because
21 4 Simpson Conventional Chromatography 215
the separation is based on the changing eluent strength rather than many Recently Felinger and Guiochon (24) compared the maximum produc-
exchanges between eluent and matrix over the length of a long column, and tion rates of overloaded elution chromatography with those of displacement
good reproducibility. The disadvantages are that it requires reequilibration chromatography for the separation of small molecules. They concluded that
before the next cycle can begin, and it is rather complicated operationally. in almost all cases the elution mode was more productive. The greater
productivity is due to the much shorter cycle time of the elution mode.
C. Frontal Chromatography Unlike displacement chromatography, with elution chromatography new feed
Frontal chromatography, also known as flow-through or breakthrough can be introduced before the previous injection has exited the column. Their
chromatography, is a mode in which conditions are chosen such that the intention is to extend the analysis to cover the case for large molecules, such
desired component very quickly saturates the column and emerges from the as proteins. For the moment, one must be content to weigh the merits of
column shortly after the feed is started. Impurities, on the other hand, are each mode qualitatively with regard to the particular application at hand.
strongly retained by the adsorbent. Feeding is continued until the front of
the adsorbed impurities approaches the column outlet. At this point the feed IV. PRELIMINARY BENCH-SCALE EXPERIMENTATION
to the column is discontinued and a wash is introduced to maximize product A. Choosing the Type and Mode of Chromatography
recovery. After the wash is collected, the column is regenerated for the next
For most separations there are many types and modes of chromatography
cycle. Advantages are exceedingly high apparent capacity (capacity is limited that will work. The real question is how to pick the method that will cost the
by impurities, which are usually a small fraction of the feed, rather than by
least, can be developed fastest, and will be the easiest and most reliable to
the desired component), and simple operation-no gradients or step changes
in eluent, and unusually high recovery of desired product because it is never operate. The guiding principle for this selection should be to examine the
high-capacity mainstays (ion exchange, hydrophobic, and dye ligand
absorbed. Disadvantages are that the product stream is diluted, that feed
must be concentrated if the column is to be used efficiently, and that chromatography) first and only go with the more exotic or low-capacity
methods (affinity, gel permeation, chromatofocusing, and high performance
impurities must bind more strongly than the desired protein. chromatography) if a real need exists. There is no advantage to employing
D. Displacement Chromatography technology at the cutting edge solely for its own sake. In the vast majority
of cases, ion exchange and the other familiar forms of conventional
Displacement chromatography is a cyclical mode of operation characterized
chromatography will easily meet the requirements of the separation. The
by equilibrating the column to a weak mobile phase, loading the column to
actual procedure for screening methods is outside the scope of this chapter.
a fraction of its capacity with a feed that is strongly retained, eluting the
It is nicely presented by Scopes and others (2,4,15,25) and will not be
column by continuously introducing an even more strongly adsorbed
substance, known as the displacer, and finally regenerating the column by repeated here.
Right from the beginning, special attention should be given to working
reequilibrating it to the starting mobile phase (21). A unique feature of this
within the constraints of industrial practice. This will ensure a smooth scale-
mode is the formation of a train of rectangular-shaped product bands with
up. The number of variables that are changed on scale-up should be kept
practically no overlap that is formed under the proper conditions of column
to a minimum with exceptions agreed to by the development team before
length, displacer concentration, and feed and displacer isotherm shapes (22).
they are adopted. A checklist is a good way to address all of the variables.
The displacer for ion exchange applications is often a charged polymer,
It is helpful at this point to work with an automated lab-scale chromato-
which can be expensive, hard to remove from the column, and may give rise
graphic system such as the Pharmacia FPLC system. This allows a series of
to worrisome residues in the product.
experiments to be carried out under well controlled conditions while
Although displacement chromatography has been known for a long time,
requiring only a minimum of operator attention. Inlet and outlet column
it has received more attention lately because it offers numerous apparent pressure gages and automatic solenoid valves for flow reversal are useful
advantages over elution chromatography. Chief among these are the higher
additions to the basic FPLC configuration. This is the time to be wide-
concentration of the eluted fractions and the higher purity of the fractions
ranging in the scope of experimentation. As the process progresses toward
resulting from the elimination of tailing of one component into the other
final production, the opportunity for change becomes less and less.
(23). On the negative side is the need to identify an appropriate displacer Once a crude separation has been attained, it should be possible to
from the standpoint of the strength of its interaction with the adsorbent and
estimate the number of columns required (see Section V.C.4). Discussion
the proper concentration to feed to the column, the need to regenerate the
of the results of the calculations among the development team will point out
column, and the cost of recovering the often expensive displacer.
Simp son Conventional Chromatography 217
216
the areas in need of further development and immediately suggest some V. SIZING OF COLUMNS TO PRODUCTION NEEDS
i mprovements. A number of cycles of experimentation, estimation, discus- A. External Influences Affecting the Design
sion, and testing of improved conditions should develop the separation to the
The design of a chromatographic separation is heavily influenced by other
point that it is ready for final scale-up.
aspects of the total project. Such factors as the production capacity, selling
The lab protocol should be developed to the point that the behavior of
price of the product, required purity, and timing of market introduction are
the chromatography can be predicted reliably. The effect of excursions in
all important in providing the best design. If the product is an extremely
the major process variables should be known as well as the required number
potent human therapeutic with life-saving potential for a previously
of plates (column height) for adequate resolution. Typical chromatograms
untreatable disease, it is going to be handled very differently from a
and a complete protocol should be provided indicating the windows for the
commodity-type enzyme designed to increase chemical processing efficiency.
various fractions to be collected. The degree of separation between the Many recombinant proteins are so potent that annual production ranges
desired protein and the closest impurity of concern should be maximized and
from several grams to several hundred kilograms (26). When such small
at least a fair understanding of the factors affecting the critical separation
volumes are involved, scale-up of the chromatography step is relatively
should be in hand. straightforward. The emphasis is on moving the product through the
The choice of the mode of chromatography seldom receives the atten-
regulatory approval process as quickly as possible. If the bench-scale
tion it deserves. The most frequent choices are on-off and elution. In view separation produces a safe and effective product, simple duplication or
of its inherently low capacity, elution chromatography should be reserved for
modest scale-up is the best course. Time saved at this stage is worth much
those cases where an exacting separation is necessary and other modes have more than the lower production cost of a fully optimized chromatography
proved inadequate. Many times the on-off mode is not exploited fully, in step. On the other hand, proteins that are needed in ton quantities, or that
that only a small fraction of the capacity is actually loaded with protein. As
must compete against existing products, or whose price is constrained by the
part of the early bench work, it should be determined just how far the resin
value of the effect they produce are the ones that need careful design to
can be loaded while still retaining acceptable yield and quality. Though insure commercial success. By deciding which group the new protein falls
seldom considered initially, frontal chromatography offers a number of i nto, the appropriate development path may be chosen. Often certain types
decided advantages that may win out in the final selection process. Frontal
of chromatography can be immediately eliminated based on these consider-
chromatography should be considered in every case to see if conditions can
ations.
be found such that the impurities are retained while the product passes
Extreme purity requirements can alter the development plan. It may
through the column. become apparent that a single conventional chromatography step will not
The Importance of Yield and Purity have the power to produce a product of high enough purity. Affinity or
B.
multiple conventional chromatography will be necessary. The sooner these
Although yield and purity are the hardest attributes to predict, they cannot
factors are recognized the better will be the final design. It makes good
be neglected in the scale-up program. Any change that affects yield or purity sense to involve biochemical engineers at a very early stage of development.
will, in the majority of cases involving biotechnology, be far more important As soon as possible a scalable process should be conceptualized. This
than the capital and productivity saved by optimizing the physical variables
focuses development work on separation techniques that can be scaled up to
of the separation. The reason for this is the high unit cost and emphasis on a commercial process.
quality characteristic of therapeutic proteins.
While yields may theoretically approach 100%, in practice they seldom B. General Approach
do. The reasons for the shortfall are not well understood. They are often Two approaches are available in scaling up chromatography. The first relies
ascribed to such catchalls as nonspecific adsorption or denaturation of the
heavily on mathematical analysis involving the solution of partial differential
protein on the adsorbent. Whatever the cause, the best course is to include equations. It attempts to discover the controlling mechanism for each part
yield and quality as responses in every experiment. One can expect differ- of the cycle and to construct a valid mathematical model. The model is
ences among various brands of the same type of adsorbent. Likewise, the manipulated to arrive at a suitable design (27). The loading phase has been
various modes of chromatography will exhibit different patterns of behavior.
218 Simpson
well analyzed in the literature. Regeneration, elution, and washing, however,

have received little attention as has the operation considered as a whole.
Extensive mathematical analysis and experimentation are necessary to model
the process adequately. Although some factors can be definitely ruled out
by this approach, in many cases the results are ambiguous. The second
approach is a more empirical and practical one that focuses on the total
cycle and progresses naturally out of early and continuing development work.
Column bed height, protein and buffer concentration, and adsorbent particle
size and brand are kept constant with the idea of simplifying the analysis and
minimizing the chance of significant scale-up effects. As a result, the design
process is relatively uncomplicated, involves only algebraic equations, and the
The numerator in Eq. (2) is the desired production rate of protein, specified
effects of each portion of the cycle can be seen in a single unifying equation. by the constant Q:
As a rule, the scale-up factors for protein separations are much smaller
than those typical of the petrochemical and chemical industries. Even high-
(3)
volume products, such as proteins from sweet whey or albumin from human
plasma, do not require columns over 700 to 2000 liters in volume (1,28).
Consequently, it is not normally necessary to change bed height or adsorbent
Its value will depend on market forecasts, the number of working days per
particle size to meet the practical requirements of large scale operation. year, and the fraction of time the columns are on-line. Note that this is the
Simply put, protein columns will not be 100 ft. tall and 30 ft. i n diameter. final production rate, not the rate at the chromatography step. (See the
Perhaps the best solution is a preliminary design based on the second
Appendix at the end of the chapter for the definition and units of symbols.)
approach followed by a review of the crucial parts of the design with the
2. Column Productivity per Cycle
more fundamental analytical techniques. In this way attention is focused on
the process as a whole and the most important parts of it. If the separation The first term of the denominator of Eq. (2), the column productivity per
is very exacting it is worthwhile, as a final check, to make sure that the best cycle, is expressed in terms of other known variables as given in equation (4).
chemistry for the separation has been chosen. Many times the answer to a
separation problem is improved chemistry rather than refinements in physical
conditions or timing of an inferior process scheme.
C. Formulation of the Design Problem

The design problem is to determine the number and size of columns needed
to produce economically a desired protein at a specified rate and purity. It
is assumed that the chromatographic separation has been established at the
bench scale. The problem can be broken down into two parts. The first part
concerns relating the number of columns to production demand, column where D = column diameter in cm, h = bed height in cm, and U = fraction
productivity per cycle, and the number of column cycles per day (29). The of the column saturation capacity that will be loaded on each cycle.
analysis is developed for on-off chromatography with indications as to how Fifty percent utilization is a reasonable figure in the absence of known
the results can be applied to elution, gel permeation, and frontal chromatog- adverse or beneficial effects of loading on yield or quality (13). The column
raphy. The second part, treated later in this chapter, is concerned with the saturation capacity C must be determined under flow and mobile phase
mechanical design of the column and choosing the best operating conditions. conditions that are identical to the proposed process. This quantity should
be determined experimentally. Care should be taken to ensure that the
1. Basic Design Equation measured capacity truly reflects the dynamic capacity of the resin. The feed
We start by expressing the production rate in terms of three factors. The should be identical to that of the full-scale process. Reconstituted pure
total production rate is the product of the number of columns, the number protein will not contain the same proteinaceous and ionic impurities as the
of cycles per day, and the output of a single column in a single cycle. genuine feed. The flow velocity must match the anticipated value, otherwise
Simpson Conventional Chromatography 221
220
While the volumes of the various column feeds depend on the size of the
problems are likely. As flow velocity increases, dynamic capacity decreases
due to kinetic and mass transfer effects. For the same reason, batch column, bed volumes are independent of the scale of operation. Expressing
adsorption tests are not as useful as those conducted in a flow system. each of the volumes processed in terms of bed volumes leads to
For this analysis, the capacity is expressed in terms of the final product
adsorbed per liter of resin. This establishes the capacity on two well-defined
quantities: the weight of final product and the volume of the resin. The
only drawback to this convention is the need to include yield estimates for
the chromatography and all subsequent processing steps. These yields, For many cases VL = VW = VE = VR = V i.e., the flow velocity through the
however, are often high and well-defined. Chromatography is usually one of column is constant over the various steps. If it is not, an equivalent bed
the last process steps and produces essentially pure protein. The other volume may be defined as
option-basing the capacity on an analysis of the feed stream-often leads
to confusing inconsistencies and high variability: impure protein solutions (12)
are notoriously hard to characterize with a single assay.
3. Column Cycles per Day where i = subscript L, W, E, or R. The velocity at the loading step is taken
The second term of the denominator of Eq. (2) is simply the working hours as the reference value. For simplicity's sake, the velocity will be taken as
per day divided by the time to complete a single cycle. Continuous 24 h/day constant for the remainder of this analysis. Hence
operation is assumed.
(6)
Bw B E , and B R are constants specific to each process. If the wash step is a

Column chromatography is a serial operation. The total cycle time is the simple displacement, 2 to 5 bed volumes will usually suffice. Elution often
sum of the loading, washing, elution, and regeneration times, as in Eq. (7). requires more volume, say 3 to 7 bed volumes in normal cases, to accommo-
date peak tailing and to maximize yield. For large-scale work step elution
(7)
patterns are favored over gradients (30). The regeneration volume is the
(8) sum of each of the volumes of regeneration and equilibration buffers, e.g.,
strong salt solution, sanitizing agent, strong buffer, and start buffer.
Normally, between 6 and 12 bed volumes are required. The final term TH
allows for a constant hold period in a sanitizing agent at zero flow. The
where L = liters processed at that step and V_ = superficial velocity through
loading bed volume is a variable and depends on the potency of the feed
the bed, cm/h. (Superficial velocity is the linear flow velocity through the stream and the working capacity of the bed.
column, assuming it is empty.) The concept of bed volumes leads to useful
simplifications at this point. A bed volume is the total of the resin and
interstitial volumes occupied by the column packing. The loading, washing,
eluting, and regenerating volumes can be represented in terms of dimension-
(15)
where P is the protein concentration in the feed in grams of final protein per
liter. Substituting Eq. (15) into equation (13) gives
222
4. Final Design Equation

All of the necessary factors in equation (2) can now be expressed in terms
of knowns to calculate the required number of columns.
Figure 1 The effect of feed potency on the number of columns required to purify
a set amount of protein. The vertex point is given by Eq. (21).
At low feed potency the number of columns is independent of bed height

5. Limiting Cases and capacity. The column's primary role has changed from separation to
Two limiting cases are of interest. The first is the situation where TH the
concentration. If the number of columns given by Eq. (18) is plotted as a
hold time, is zero or negligible.
function of feed concentration, the curve in Fig. 1 is obtained. Concentrating
the feed to a potency to the right of the curve's "knee" will significantly
reduce the number of columns required and the total expense. The location
of the "knee" is the vertex of the hyperbolic curve given by
(21)
In this case the number of columns is independent of the bed height. A
shallow bed cycling frequently is equivalent to a deep bed cycling slowly.
The real factors that determine bed height are the desired resolution, cycle
ti me, and pressure drop. The trend in industry is toward relatively short bed 6. Application to Other Modes of Chromatography
heights, say 15 to 45 cm. Classical chromatography theory predicts that
resolution should be directly proportional to the square root of the bed Elution chromatography is distinguished by a very low value of U, the
fractional approach to the total capacity. This corresponds to very light
height (31); however, changes in buffer pH, ionic strength, gradient shape,
or choice of resin are far more flexible and attractive ways to improve loading of the column with a small volume of relatively concentrated protein.
The exact amount to be loaded is determined by lab or preparative scale
resolution. A fuller discussion on the proper choice of bed height is tests. There is no washing or regeneration step. Gradient elution requires
addressed later in this chapter.
The second limiting case is the situation where the loading step an equilibration step to ready the column for the next cycle. Isocratic elution
requires no equilibration. In isocratic cases it may be possible to overlap the
dominates all others. This is encountered when the potency of the feed
ti me cycles of successive batches to some degree (32).
stream is very low. Eq. (18) reduces to
Gel permeation chromatography is amenable to this analysis if U is set
(20) equal to one and C is set equal to 5 to 15% of the feed potency (8). A
highly concentrated feed and much greater capacity are possible if desalting
is the objective (8). There is no washing or regeneration step. Elution of
the lowest-molecular-weight components is usually finished by about one bed
volume. The highest-molecular-weight components elute no sooner than the

void volume of the column, about 0.35 bed volumes. The fact that GPC has
no regeneration step allows the next feed cycle to start before the previous
batch has completely eluted. A decrease to about 0.6 bed volume for the
elution volume (apparent) is possible. The scale-up of GPC has been treated
by a number of authors (8,10). These references should be consulted for a
fuller discussion of the unique requirements of GPC.
Frontal chromatography can be handled with the same equations if the
capacity of the resin is specified in the correct way. In this case the capacity
C is defined as the weight in grams of desired protein that flows through a
one-liter column before the breakthrough of retained impurity reaches an
unacceptable level. In this mode of operation, U is usually close to unity.
Naturally, the elution time and volume are zero.
D. Determining the Rate Controlling Step

The adsorption process can be broken down into a number of steps that take
place in series. As shown in Fig. 2, the protein dissolved in the bulk liquid
must diffuse across a thin, stagnant liquid layer surrounding the outside of
the adsorbent particle. Once at the surface of the particle, the protein must
diffuse into a pore leading to the interior. Diffusion continues until an unoc-
cupied adsorption site is encountered. The protein then adsorbs to the site
at some finite rate. In most cases one step of this sequence is much slower
than the others. This step is known as the rate determining or rate
controlling step. It sets the overall adsorption rate because of the serial Figure 2 Pictorial representation of steps in the adsorption of protein. (1) Protein
nature of the process. in well-mixed bulk liquid approaches adsorbent particle; (2) protein diffuses through
A closer examination of the rate controlling step in the chromatography thin film surrounding adsorbent particle; (3) protein diffuses through tortuous pore
is in order once the number of columns has been estimated. With this toward adsorbent site; (4) kinetic adsorption step; (5) adsorbed protein; (6) unoc-
knowledge one can gain a deeper understanding of the factors that influence cupied adsorption site.
the column productivity. Changes that are inevitably suggested by plant
personnel can be reviewed more intelligently. Those that might affect the ment, numerous assumptions have to be made in order to make the problem
controlling mechanism deserve a closer look, while those that are predicted tractable. It is assumed that the contribution of axial dispersion is small
to have no effect can be adopted with minimal testing. One should compared to the effects of pore diffusion, film resistance, or binding kinetics.
remember that any changes that improve the controlling rate might switch Adsorption is taken to be essentially irreversible, and equilibrium is assumed
control to a different step. Furthermore, the rate controlling step will likely between pore liquid and the internal surface of the adsorbent.
change with the switch from adsorption, to wash, to elution, and regenera- Finally, the concentration profiles are assumed to approach a constant
tion. shape as they make their way through the column. If all of these assump-
Over the years there have been many attempts to model chromatograph- tions are satisfied, the breakthrough curve will depend solely on the rate
ic operations. They have ranged from simple lumped parameter approaches determining step. If these assumptions are not satisfied, other appropriate
to numerical integration of complex, coupled partial differential equations. solutions found in the literature should be applied. Another approach is to
Rather than review each of these solutions, one set that covers many model the system with a set of computer programs such as those being
biotechnological applications has been chosen for emphasis. currently developed by Sweetenham and coworkers (37) in Great Britain.
The papers of Arnold, Blanche, and Wilke (33,34,35,36) give a Pore diffusion, fluid film mass transfer, axial dispersion, and slow binding
convenient and systematic way to determine the rate controlling step for kinetics are considered as possible candidates for the rate controlling step.
chromatography operated in the on-off mode. As with any analytical treat-
The following expressions give thnumber of transfer units for each of the it exits the column along with a measure of the peak spreading can be used
possibilities. to deduce the adsorption isotherm and effective diffusivity for transport
within the pores. What is the likely rate determining step for adsorption in
most conventional protein chromatography? In all but extremely short beds,
axial dispersion will be negligible. Restriction due to slow mass transfer
across the stagnant fluid film surrounding each particle is seldom a problem
either, unless the feed is very dilute. The usual rate determining step is
diffusion through the pores of the adsorbent (33). For such cases, the
breakthrough curve for adsorption is given by
(28)
If the particles are extremely small or the protein-adsorbent interaction

involves large molecules with exacting steric requirements, the possibility
exists that binding kinetics may become dominant. Likewise, binding kinetics
may control in the elution phase of the chromatography. Solutions are given
by Arnold (33) for the various other rate controlling steps.
Once the rate determining step has been tentatively identified, it should
The rate determining: step is the one that gives the least number of transfer be confirmed by further experimentation. Varying the size of the adsorbent
units when evaluated - with the data specific to the situation under study. Two particles is a good way to confirm the controlling step. If the response in N,
of the equations (Nf and Nd) car be evaluated with the aid of correlations. the number of transfer units, is very strong, it is likely that pore diffusion
The mass transfer correlation of Foo and Rice (38) is appropriate for the controls (N is proportional to 1/R'). If the effect is moderate, stagnant film
estimation of the fluid film masstransfer coefficient. mass transfer or axial dispersion are confirmed (N is proportional to 1/R).
Lack of response would confirm slow binding kinetics.
(26) Activation energies derived from temperature studies can aid in
discriminating between physical mass transfer limitations (activation energies
of 3 to 5 Kcal) and chemical reaction control (activation energies greater
than 10 Kcal). None of these tests should be taken singly as proof of the
For uniform spheres, the speific external area is given by ap = 3(1-E)/R. identity of the rate controlling step. All available data must be considered
The diffusion coefficient of the protein in a free liquid can be estimated from as a whole to arrive at the best choice. In many real-life cases the data will
(39) not be clear-cut and will require careful interpretation.
(27) E. Effect of Process Variables

1. Nature and Concentration of the Feed Stream
The composition and quantity of the feed stream can have a significant effect
on the final design. Although the purpose of chromatography is purification,
or handbook data can be used. The Peclet number can be estimated from the job is made much easier the purer the feed stream is. Industrial
the work of Miller and King (40), who showed that the product EPe chromatography must be robust, economical, and effective. To insure that
approaches a constanrt value of about 0.2 at low Reynolds number in packed these objectives are met, the following questions concerning the nature of the
beds. The number of transfer units for pore diffusion or binding kinetics can feed should be answered.
only be estimated with the help oflata from experiments. Column runs with Is the feed in the right composition range? If it is too concentrated,
pulse inputs of different concentrations and at different flow rates will give density differences, inadequate buffering capacity, and high viscosity may
the desired information (34). Measurement of the position of the pulse as cause problems. If it is too dilute, the column will be occupied all the time
22 8
just adsorbing the product. For GPC the limits of protein concentration are found at the low end. Smaller-scale operations based on freely permeable
10 g/L to 70 g/L depending on the degree of resolution required (41). adsorbent particles (44), membrane chromatography, or small-diameter
Simple desalting allows operation at the high end of the concentration range columns (particles given extra support by proximity to the wall) can be
as well as increases in the volume of feed per cycle. For ion exchange the operated at velocities up to 5000 cm/h. The kinetics of adsorption can
protein of interest should be at a concentration of 5 to 10 g/L with a total become a limiting factor under these conditions.
protein concentration no greater than 25 g/L (2). Solubility limits may lower The effect of flow velocity on column efficiency is commonly illustrated
these maxima. In no case should the viscosity of the feed be greater than with the Van Deemter (45) equation
twice the eluent viscosity. This corresponds to a protein concentration of
about 70 g/L in a dilute aqueous buffer (13,42).
Does the feed need conditioning, e.g., filtration or a change in ionic
strength or pH? Adsorbent beds are very efficient filters. Feeds to the HETP or the height equivalent of a theoretical plate is directly related to the
column must be no more than very slightly hazy, otherwise the column may efficiency of the column. The smaller the HETP the greater the number of
plug with solids. Aside from GPC, chromatography requires exacting control separation stages available to resolve mixtures. The plate height is
of ionic strength and pH. Even for GPC, the ionic strength should be above influenced by three factors. The first term, a„ accounts for undesired mixing
0.15 if hydrohobic interaction is to be minimized (43). The higher feed due to imperfections in packing, liquid maldistribution, and dead volume in
concentrations typical of industrial chromatography often require increases valves and lines. The second term, a2u0, is directly proportional to the
i n buffer strength compared to the original lab procedure. The buffer pK superficial velocity. It depends on the rate of mass transfer from the bulk
should be within 0.5 of the pH of the eluent. The feed must be brought to liquid to the surface and interior of the adsorbent. It includes both the thin-
the conductivity and pH that were determined to be optimum in the lab. fil m mass transfer resistance at the surface of the adsorbent and the intrapar-
This requirement may prompt a change to a new sequence of process steps ticle resistance due to diffusion within the particle. The final term, a3/u0, is
that minimizes transfers from one buffer to another. proportional to the ratio of velocity of free diffusion in the direction of flow
Has the cleanest starting material been chosen? If not, fouling may be to the velocity due to convected flow. Macromolecular diffusion coefficients
a problem and capacity will suffer. For crude feeds, a small guard column are so small that a3/u0 can be neglected in virtually all industrial applications.
may be beneficial, or the chromatography might be better put off to a step The net result for the HETP analysis for practical cases is that plate
later in the process. height continually increases with increasing flow velocity. Mechanical design
The answers to all these questions depend to a great degree on the of the column system should be aimed at minimizing dead volume and
specific protein and process under review. A balance must be struck between maldistribution. The slope of the increase in HETP can be controlled by
the expense of chromatography and the expense of pretreatment or other choosing adsorbents with properly tailored internal structures that do not
alternatives. Generally, pretreatment will be less expensive. impede the transport of the protein.
2. Flow Velocity 3. Temperature
The number of columns required to produce a set quantity of protein is Temperature can easily indirectly affect product purity or recovery. Often
directly related to the flow velocity through the columns. Velocity should be lab procedures will be carried out at 4°C just as a routine precaution against
increased to the point that the upper pressure drop limit is approached or degradation of the protein. It is costly, however, to maintain 4°C throughout
resolution deteriorates to an unacceptable degree. Flow velocity will scale-up. Besides the obvious cost of cooling the column, there is the cost
influence the rate of mass transfer from the bulk liquid to the outer surface of air conditioning the surrounding room to prevent sweating of the cold
of the adsorbent. Higher velocity also increases the extent of eddy diffusion equipment, the cost of insulation, and the cost to cool all the buffers, eluents,
and backmixing. In extreme cases the flow can be increased to the point that and regenerants. Additional factors are higher pressure drop due to
the protein in the feed will be swept out of the column before it has a chance increased liquid viscosity, and possible changes in the behavior of the
to adsorb. adsorbent due to increased importance of hydrophobic forces at low
Superficial velocities normally range from a low of 10-20 cm/h up to a temperature (46). Optimized, quick operation at room temperature will
high of about 400 cm/h for large-scale applications (19). Low-molecular- often give the same result as direct scale-up of a lengthy, preliminary lab
weight peptides and proteins and rigid adsorbents favor the high-velocity process at 4°C.
range. Very-high-molecular-weight proteins and compressible adsorbents are
Conventional Chromatography 231
The upper temperature limit of operation is set by the stability of the adsorbent to the same height as will be used in the plant. A small-diameter
protein and the yield of the chromatography step. Diffusional processes are column is suitable as long as the ratio of adsorbent particle diameter to
less of a problem at higher temperatures (47), but whether these effects are column diameter is greater than 30. The bed is equilibrated to the eluent
significant enough to warrant raising the temperature above ambient in buffer. Next, feed of the same composition and history as that anticipated
production units can only be determined by experiments in the pilot plant. in production is pumped to the column at a specified superficial velocity in
The main point is to look at the effect of temperature closely and not the range of 40 to 400 cm/h. The effluent from the column is continuously
follow the general rule of thumb that colder is better in all circumstances. monitored for ultraviolet light absorption at a suitable wavelength, usually
4. Properties of the Adsorbent 280 nm. A curve similar to Fig. 3 should be obtained. If time zero is taken
as the start of the feed, the curve may be broken up into five regions. The
a. Type of Adsorbent. There is a bewildering array of adsorbents on the
first is the portion covered by the time for the front of the feed to make its
market. The choice of a few candidates for further study can be guided by way through the bed. During this time the UV trace is unchanged from the
the discussion at the beginning of this chapter. In many cases, several types buffer equilibration level. When the front of the feed reaches the end of the
of chromatography can yield essentially equivalent results in terms of
column there will usually be a small, rapid increase in UV response corre-
separation and cost. At this stage what is desired is to eliminate those types sponding to the breakthrough of unadsorbed minor impurities in the feed.
that are clearly not suited to the separation, rather than make a definitive
This is followed by region of shallowly increasing UV activity due to start of
choice of the "best." If two or more chromatography steps are involved in breakthrough of the desired component. At some point the UV trace starts
the process, it is a good idea to choose types that differ in their basis of
to increase rapidly due to impending saturation of the adsorbent capacity.
separation. For example, one might choose hydrophobic chromatography This sigmoid portion of the curve is followed by an asymptotic approach to
(hydrophobic interaction) to follow anion exchange (electostatic interaction)
the UV activity of the feed as the column is completely saturated. Lower
in preference to two anion exchange steps.
superficial velocities give breakthrough curves approaching a step function.
Once several candidate types of chromatography have been identified,
extensive lab experimentation should be done to develop a lab-scale
separation process. Once the type of chromatography has been specified,
choose three or four representative brands of adsorbent within that group for
testing, because such factors as protein recovery and recyclability cannot be
predicted. There is normally no need to test every available brand exhaus-
tively. Other factors to consider in picking adsorbents include cost,
availability in bulk, thoroughness of documented manufacturing and quality
control methods suitable for presentation to regulatory authorities, location
of the source, customer service, and history of quality control.
b. Capacity of the Adsorbent. Capacity has a direct and strong effect on the
number of columns required to purify a protein. Consequently, it should be
as high as possible consistent with good recovery. What matters is dynamic
capacity rather than thermodynamic capacity. The capacity to adsorb the
protein of interest must be determined under flow and buffer conditions
identical to those specified for production (25). Capacities determined from
acid-base titrations or other indirect methods should be taken as only
suggestive of the dynamic capacity. In the preliminary design phase, the
correlation of Scopes (2) is a useful tool to estimate the approximate ion
exchange capacity for a protein of known molecular weight. For the final Figure 3 Breakthrough curve. (1) Dead time for feed to transit column assuming
design, there is no substitute for experimental determination of the capacity
no adsorption; (2) breakthrough of unadsorbed impurities; (3) start of breakthrough
employing the exact feed and flow conditions of the design. of desired protein; (4) main portion of breakthrough; (5) final approach to total
The determination of the breakthrough curve is the best way to saturation.
determine the dynamic capacity. A column is packed with the production
232 Simp son Conventional Chromatography 233
Higher velocities give progressively shallower curves with lower dynamic Decreased diffusional path means less peak tailing and increased flow
capacities. The shallow curves reflect increasing mass transfer resistance. velocity before additional resistances come into play. Assuming an upper
Capacity can be defined in a number of ways. The first is taken to be pressure drop of 50 to 100 psig and more typically 10 to 50 psig as a working
the grams of protein fed to the column at the point that the UV trace value for aqueous solutions, the minimum particle size range is from 50 to
reaches a certain percentage of the final UV level, usually 5%. A variant of 100 microns at bed depths of 15 to 60 cm. Pressure drop through an
this would be to choose the point that the integrated loss of desired product incompressible packed bed of solids can be predicted by the equation of
reaches a set value, again usually 5%. The number of grams of protein is Ergun (49) modified for laminar flow
divided by the bed volume in liters to obtain the specific capacity. Further
(30)
adjustments for column and subsequent processing yields convert the capacity
to the basis of grams of final protein per liter of adsorbent. The other
popular definition is based on the final approach to saturation: either the
50% or the 95% level of the final UV activity. While perfectly legitimate, Note that particle diameter appears in the denominator and is raised to the
these latter definitions give capacities that are never approached in normal second power. Halving the particle size quadruples the pressure drop.
industrial practice. For this reason they are misleading, and their use should Given that most gels on the market already are optimized to the 10 to 50
psig pressure drop at fast flow rates, this does not leave a lot of room to play
be avoided.
Dynamic capacity reflects the effects of mass transfer limitations as well with. Recently there has been a trend toward 30 to 50 micron particles for
as the ultimate capacity in the thermodynamic sense. Mass transfer limita- small-volume applications where the increased resolving power compensates
tions occur as the result of the interaction of flow velocity, particle size, pore for the higher pressure drop (50). The bed void volume is influenced by the
size and pore size distribution, and the physical properties of the protein, tightness of the particle size distribution and the quality of the packing (13,31,
buffer, and adsorbent. 51). Very small changes in the void volume are rapidly reflected in changes
Capacity is directly related to the accessible surface area of the gel and in the bed pressure drop. Some manufacturers will supply, at extra cost,
the density of adsorbent sites on the surface. If the ligand at the adsorbent adsorbent with a narrow particle size distribution (52). The uniformity of
site is small, it is possible to crowd too high a density of sites on the surface particle size makes for sharper bands and greater resolving power. Viscosity
of the gel. If the protein interacts with too many sites, it will be very difficult of the feed and eluent also is directly reflected in increased pressure drop.
to displace. With large ligands, such as monoclonal antibodies, it is hard to Choice of solvent, concentration, and temperature of operation should be
attach enough in the right configuration to give high capacity (20). Non- reviewed with this in mind. For dilute aqueous solutions, one can expect
porous gel beads with adsorbent sites just on the surface (pellicular) have about twice the pressure drop at 4°C as would be obtained at 24°C (8).
very low capacities (48). These gels, however, do have very fast mass transfer d. Particle Shape. Spherical beads should be favored over irregularly shaped
characteristics, which makes them popular for analytical applications. particles on the basis of greater permeability in packed beds and better bed
Porosity increases surface area and ultimate capacity dramatically, especially stability on repeated use (51). Rough particles do have a price advantage
if the pores have a small diameter. There is a limit, however, to the degree over smooth spheres, which may warrant consideration in cases where the gel
of porosity that is practically attainable. Too high a porosity leads to poor is used just once. Particles shaped like needles offer the advantage of faster
mechanical strength of the particles. Arbitrarily decreasing the pore size adsorption kinetics by way of reduced diffusional distance compared to
ultimately leads to exclusion of the protein from the interior of the particle spheres of equivalent volume, but they raise concerns about packing and
(see Section F below for a fuller discussion of pore size). pressure drop.
Capacity is often a strong function of pH, ionic strength, and polarity of e. Compressibility and Swelling of the Adsorbent. When protein chromatog-
raphy first developed, many of the adsorbents available were easily
the eluent. Concerns such as these affect the choice of strong vs. weak ion
exchangers, the strength and pH of buffers, and the proper location in the compressed by very modest pressure gradients (53). In the most extreme
process of the chromatography step. cases, just the weight of the bed was enough to deform the particles on the
c. Particle Size, Particle Size Distribution, and Shape of the Adsorbent. In bottom of the bed to the point that they completely blocked flow. The gels
general the particle should be as small as possible subject to pressure drop also dramatically expanded and contracted with changes in eluent composi-
constraints and packing difficulties. Small particle diameter minimizes tion, which led to the formation of voids and channels in the bed (16).
intraparticle diffusional resistances, which account in most practical cases for Manufacturers of adsorbents have been very aware of these shortcomings
most of the mass transfer limitation of production scale chromatography. and have introduced quite a number of improved adsorbents that hardly
The exponential term makes it clear that operation beyond a critical pressure
drop per unit length of bed can lead to complete blockage of the column.
Comparison of this equation to Eq. (30) shows that they are equal except for
the last factor. The exponential term reflects a dependence of void volume
on pressure as a result of deformation of the particles. Hysteresis and very
slow recovery from high pressure drop excursion are typical of compressible
adsorbents.
f. Porosity, Pore Size, and Pore Size Distribution. Since proteins are very
large molecules, the issue of entry into the interior of the adsorbent cannot
be overlooked. Pores should be large enough that free diffusion is not
hindered significantly. This normally can be met with pores that are 10 to
20 times the diameter of the protein (56). Regnier (57) recommends 100
angstrom pores for small peptides and proteins, 300 angstrom pores for
50,000 to 100,000 dalton proteins, and 500 to 1000 angstrom pores for
macromolecules larger than 200,000 daltons. The approximate diameter of
globular proteins can be estimated from the equation shown below, which is
adapted from Creighton (46).
Alternatively, the manufacturer's tests of protein exclusion can be consulted

to choose the appropriate gel. If pore size is increased indefinitely,
mechanical strength will suffer and capacity will fall. The freedom to tailor
Figure 4 Pressure-flow rate relation for DEAE-Sepharose Fast Flow at a constant the distribution of pore sizes has been exploited recently with the introduc-
bed height of 15 cm and different column diameters. (Adapted from Groundwater, tion of an ion exchange resin consisting of 10 to 20 micron particles with a
Ref. 131.) bimodal pore size distribution (44). The larger pores at 6000-8000
angstroms are said to allow convective flow through the particle to some
compress under high flows or shrink or swell with changes in eluent extent. The smaller pores at 500-1500 angstrom branch off the larger pores
composition. Only in gel filtration of very-high-molecular-weight compounds and contain most of the sites for adsorption. Concentration gradients within
is the problem still evident. In this case the gel has to be very open with the particle are much reduced compared to conventional gels. High dynamic
large pores, which adversely affects its mechanical strength (54). Supports capacity is claimed at superficial velocities of up to 5000 cm/h.
that are derived from natural products tend to be more compressible than g. Stability Toward Chemical and Microbial Attack. The gel under
those developed on inorganic bases. In small columns, wall effects definitely consideration should be stable to all the reagents that will contact it during
reduce compressibility and the associated pressure drop. The effect can be the chromatographic cycle. All commercial gels are stable in aqueous media
felt even up to a bed diameter of 37 cm (see Fig. 4). The pressure drop at slightly acidic to neutral pH. Most are stable to the common buffering
relationship in the case of compressible beds is given by Janson and Hedman salts and chaotropic agents such as guanidine hydrochloride, urea, and
(55). sodium dodecyl sulfate. Some, however, have difficulty with extremes of pH
23 6
that may be required in the sanitizing and depyrogenation steps of the cycle. F. Optimization of Individual Steps
Sodium hydroxide has so many advantages as a cleansing, sanitizing, and Chromatography is a cyclical operation. To prepare the best design, all steps
depyrogenating agent (16) that short-term resistance of the resin to 0.5 N of the cycle must be considered. The general cycle of operation consists of
NaOH and long-term resistance to 0.1 N NaOH are very desirable proper- loading the column with feed solution, washing the residual feed off the
ties. Cross-linked agarose will pass these tests. Silica based adsorbents column, eluting the column, and finally regenerating and sanitizing the
(25,58) and cellulosic supports dissolve under these conditions. At times it column preparatory to the next cycle. Certain types of chromatography are
may be necessary to treat the adsorbent with sodium phosphate, detergents, simpler than the cycle given above.
alcohol, or acetic acid in order to remove strongly adsorbed impurities The stepwise nature of chromatography offers opportunities for
(16,58). Stability toward these agents should be checked. optimizing each of the steps separately. For instance, in the equilibration
If the resin slowly degrades, it is important to determine at what rate and regeneration steps much smaller molecules with faster transfer kinetics
and whether the degradation products are toxic. Mechanical strength of the are being adsorbed and desorbed. This means that flow rates can be
gel may also suffer, leading to compression or the formation of fines. considerably higher than those specified for protein adsorption and elution.
Breakdown of the attachment of the adsorbent ligand to the supporting gel The faster equilibration and regeneration translates into faster turnaround
or oxidation of the ligand may be a problem. All adsorbents should be ti me and fewer columns.
freshly regenerated before use to be sure any breakdown products formed Flow direction may be changed if it is desired to remove very strongly
during long-term storage are washed from the gel. bound impurities in a small volume and time (20). If impurities are bound
Microbes will proliferate on most gels if nothing is done to suppress strongly to the top of the column (assuming loading in downflow), regenerat
growth. The severity of the problem depends on the conditions of the ing upflow will sweep the impurities off the column without having to push
chromatography and the incoming bioburden. Physiological pH and dilute them all the way through the bed.
phosphate buffers at room temperature favor growth (6). Low temperature, It may also be possible entirely or intermittently to skip a sanitization
extreme pH (less than 3 and greater than 10), and mildly disinfecting buffers, cycle in cases of light fouling of the gel. Reequilibration can be started with
such as borate, make it less likely. Feeds from earlier in the separation high concentrations of buffer to take advantage of mass action to speed
process tend to be more of a problem than those close to the end. The things up (19). The equilibration is finished with a relatively small volume
crude feeds usually have a higher bioburden and more nutrients for growth. of buffer whose strength matches that of the column feed.
In some cases the gel itself may serve as the food source (31). In others the The temperature of each step can be adjusted, but given the high
adsorbent particles just serve as attachment sites. thermal capacity of the column, usually only one step will be considered for
h. Adsorbent Lifetime. The question of adsorbent lifetime is one for which change. A good candidate is the sanitization step. The effectiveness of the
there is no easy answer. Aside from the obvious extended use test, there is cleansing, sanitizing, or sterilizing agent is greatly increased by raising the
no way to predict the lifetime of a gel in a particular separation. What can temperature (1,30), thereby shortening the column turnaround time by hours.
be done easily is to perform a sensitivity analysis of the effect of gel lifetime
on the economics of the process. If a small change in the estimated gel VI. MECHANICAL DESIGN AND OPERATION
lifetime translates to a large cost increase, there is plenty of incentive to go
ahead with the time trial or find a cheaper gel. If the effect is small, the A. Column Geometry
pilot experience as it accumulates during development will suffice as an Columns are available in diameters ranging from 0.5 cm up to 2 m and in
estimate of the gel lifetime. The data of Viljoen et al. (59) are illustrative l engths from 5 cm up to 1 m or more (61). Ion exchange columns of several
of the approximate lifetime that has been reported in the literature. They thousand liters bed volume are in use today (13). For protein work, other
report lifetimes of over 1000 regenerations for the purification of human than gel permeation, rather short columns of from 15 to 45 cm are favored,
albumin on a production scale. Total protein fed to the column may be a principally because the adsorption-desorption behavior of the protein negates
better gage of life expectancy than the number of regenerations or elapsed almost all of the benefit of long columns, while the debits of high pressure
time (60). Fouling is usually a more important determinant of lifetime than drop and greater compression of the gel remain (20). The trend toward
chemical stability of the matrix (4). Estimates in the literature should be short squat columns means that excellent liquid distribution is a must.
taken as nothing more than an indication of what is possible in some Vendors have gone to considerable effort to design liquid distributors that
circumstances. There is no substitute for actual experience. direct the flow evenly over the face of the column and minimize dead volume
(31). They generally employ tortuous path, antijetting devices and stacked
screens to insure proper operation. Two designs are shown in Figs. 5 and 6.
In large diameter columns a fine screen over a coarse screen gives more
reliable distribution than a porous plastic disk (31).
Fixed or movable head designs are available. The fixed head is less
expensive, and easier to seal, but it is harder to pack and not as versatile as
the movable head design. With the movable head, either one or both ends
of the column are adjustable, so that the distributor can be brought down to
the surface of the bed to eliminate dead volume.
O-ring seals prevent leakage past the movable distributor. In larger
diameters, the problems of seal leakage, ease of adjustment, and parallel
lowering of the distributor become more difficult. Newer designs address
some of these problems.
Material of construction is usually glass or clear plastic for small
columns (lab and pilot scale) and stainless steel or fluorocarbon-coated
stainless steel for production columns. The plastic and glass designs are
fragile and are limited to around 15 psig operating pressure in larger
diameters (62,63). They do offer the advantage of seeing the bed at all
times. Thicker wall designs are available on special order for higher pressure
Figure 6 Amicon design of flow distributor for chromatography column.

(Reprinted with permission from Amicon Division, W. R. Grace and Co.)
operation of large industrial columns (64). Adsorption of seemingly colorless

protein solutions can cause a subtle change in the appearance of the gel that
can be detected by the naked eye. Fissures or other gross abnormalities are
easily spotted. Due to irregularities in concentricity, very large diameter
glass columns are custom fitted to the distributor. If the glass wall of the
column breaks, new end plates are needed in addition to a new glass tube.
Plastic suffers from gradual darkening and crazing due to action of light and
the aggressive chemicals used for sanitization and regeneration. If the
crazing continues to the point of cracking the wall, sudden failure of the
Figure 5 Cross-section of the upper part of a stainless steel Sephamatic Gel Filter column is possible.
belonging to the 08-series with a diameter of 80 cm. (1) Feed manifold of four pipes; The stainless steel units are rugged and can operate at 45-60 psig or
(2) false head cover for structural strength; (3) coarse distributor screen; (4) fine beyond with heavier construction. Stainless steel is susceptible to the
screen to retain adsorbent; (5) true head plate; (6) adsorbent bed. (Adapted from corrosive action of buffer salts, especially chlorides. The more corrosion
Janson, Ref. 67.) resistant type 316 should be specified in preference to type 304. Sanitary 3A
240
construction with electropolishing minimizes crevices that promote corrosion D. Pumps
and provides for easy cleaning. ASME or similar coded vessels are available The main requirements for pumps are that the flow be reproducible, pulse-
if desired (65). The major disadvantage of the stainless units is that they are free, and adjustable. For small lab or pilot columns, these conditions can be
opaque. The transition from a clear pilot column to an opaque production satisfied by peristaltic pumps (55). Their main advantages are the lack of
column means giving up a process control that development personnel may moving seals, ease of cleaning, and gentle pumping action. Disadvantages
have come to rely on heavily. On the other hand, large diameter columns are an upper pressure rating of 20 to 30 psig, slow drift in reproducibility of
of glass only show the condition of the bed in the vicinity of the wall. Visual the flow as the pump tubing ages, slight pulsing in the flow, and a rather low
appearance under these circumstances is not as reliable as periodic HETP upper limit to the flow capacity.
determinations (26) (see Section J below). The final decision on the material For large pilot or production columns, rotary lobe pumps, gear pumps,
of construction depends on the size of the production columns and the true or double diaphragm pumps are good choices (30). In general these pumps
usefulness of being able to observe the bed. are more rugged than the peristaltic type. They also have a better long term
If the adsorbent is very compressible and yet a long bed is needed for reproducibility of flow and can easily reach 100 psig. Flow capacity is not a
resolution, shallow 15 cm beds may be stacked one above the other (66). problem. While pulsing is still present, it is usually less pronounced than
Gel permeation chromatography of very high molecular proteins would fall that of a peristaltic pump. Disadvantages of these types of pump are that
in this category. they have moving seals, they are harder to clean, they require pressure relief
devices, and their corrosion resistance may be a problem.
B. Column Prefilters Centrifugal pumps are not recommended because they require separate
All chromatography columns should be protected from plugging by separately control systems to maintain constant flow under changing backpressure.
filtering all feeds to the column (67) The filter should retain particles smaller Under certain flow conditions they may also subject the protein to elevated
than 1/10 to 1/100 of the diameter of the adsorbent beads. Pleated depth- temperatures or slight cavitation that will probably cause aggregation or
type cartridge filters are satisfactory. The filters should be sized to offer denaturation.
negligible pressure drop and minimal dead volume. Appropriate pressure
gages should be installed to follow the condition of the filters. The prefilters E. Detectors
need to be sanitized in the same way as the column. In-line filters that can The detection scheme that is almost universally specified for proteins is
be easily disassembled are best. ultraviolet light absorption at 280 nm. The absorption is due primarily to
tryptophan and tyrosine residues with a very small contribution from
C. Tanks phenylalanine (46). Virtually any protein will contain at least some of these
If the chromatography step is close to the end of the process and represents amino acids. The situation is not as sure for peptides. If absorption at 280
the final purification step, type 316 stainless steel tanks that conform to the nm is inadequate, the wavelength can be dropped to 205 nm for a more
3A sanitary code should be specified (30). An exception to this rule is tanks universal response (68). Of course the lack of specificity at 205 nm means
for buffers that are high in chlorides or other components corrosive to i mpurities may be falsely identified as peptides.
stainless steel. Plastic is usually a good alternative for such cases, provided The detector light source is usually a long-life lamp and filter
it is impervious, nonleaching, smooth, and constructed to be free-draining. combination. Filters are available at wavelengths of 206, 226, 254, 280, 313,
For production units, each feed stream to the column should come from a 365, and 405 nm. This setup has a much longer lamp lifetime (6000 vs. 200
separate tank in order to minimize the chance of cross contamination. These h) than the deuterium lamp found in infinitely adjustable wavelength
tanks do not have to be any larger than the amount needed for a day's detectors.
operation. Large inventories of dilute buffers at mild pH's should be avoided Flow cells are available for all flows from the analytical scale to about
due to the possibility of the buildup of mold and bacteria. Stock buffer 30 Lpm (30). Flow cells with a short and long path length are handy to get
solutions may be prepared in highly concentrated form, which prevents good sensitivity for impurities while still keeping the main component on
spoilage. These concentrates are then fed from large holding tanks to the scale. Appropriate flow cells for truly large-scale production are limited.
smaller feed tanks and diluted as needed. Ideally each tank should be The approach taken by vendors has been to split the flow and only send a
equipped with agitation, a heating/cooling jacket, and load cells. small portion to the detector. While this can be made to work, it would be
much better to have the total flow pass through the detector.
Simpson Conventional Chromatography
24 2 243
Another problem with commercial separations is that the concentration Table 1 Pressure Drop and Flow Velocities for Water Flow
of protein coming off the column during elution will often saturate the at Room Temperature in Smooth Pipes at Reynolds Numbers
of 2000 to 50,000
response of the detector. Instead of Gaussian peaks the result is a peak with
a flat top. Integration of the peak to determine the amount of protein is
pointless. The total eluate has to be collected and sampled if an assay of the
protein content is desired.
It is often useful to monitor conductivity and pH in addition to UV
absorption. Various conductivity in-line flow cells are available for all scales
of operation. Conductivity is a good way to verify the composition of buffers
and monitor the progress of regeneration of ion exchangers. Units with
temperature compensation and automatic range finding should be specified.
Although pH changes are seldom used to displace proteins from absorbents Source: Adapted from Johansson (30).
(2), it is good practice to monitor the pH profile of the chromatography
cycle. The profile can become part of the process validation. In-line pH
probes are available for all scales of operation. Again temperature units. Three-way valves are also popular as a means of reducing the
compensated units are preferred. complexity of the piping involved in providing for flow in either direction
through the column. Pressure gages should be installed as close to the
F. Air Traps column inlet and outlet as possible. This insures that the pressure drop
Air traps are necessary to prevent air from entering the column and inter- reflects only the gradient across the bed. The column should be plumbed for
fering with proper flow distribution (30). The design is fairly simple. The flow in either direction by switching only a few valves.
trap is just an upright clear plastic or glass cylinder with inlet and outlet Connections to the top head plate should be easily disconnected to allow
ports in the bottom head. Feed and entrained air enter the bottom of the the column to be packed. A vent valve is needed on each line leading to the
cylinder where the air disengages and accumulates at the top. The liquid, head plates to allow for purging all the air from the lines on startup and
free of bubbles, exits through the outlet port on the bottom of the cylinder. column packing. If piping manifolds are installed, be sure any possible cross
Periodically the air accumulated at the top of the trap is released by opening contamination can be tolerated, otherwise adequate checks for leaks have to
a vent valve. The trap must be big enough that the air can disengage from be written into the operating procedure. The design should contain no dead
the liquid before the feed leaves the trap. An adequate design is to have the legs or other plumbing that cannot be fully wetted by the sanitizing agent.
trap diameter 6 to 8 times the diameter of the inlet port. The height of the H. Flowmeters
trap should be about 1.5 times the diameter. The trap is a fairly large dead
volume. In situations where it is important to minimize backmixing, the trap On the lab or small pilot plant scale, flow rates are often just checked
can be drained just before switching over to the new feed. If the flow to the periodically with graduated cylinders and a stopwatch. For larger scale
column can be reversed, design the plumbing so that the air trap is always operation, rotameters, turbine flowmeters, or Coriolis effect flowmeters are
on the feed side of the column. preferred. Johansson (30) especially favors a bearingless flowmeter with a
light reflective rotor sensed by a photodetector. Glass rotameters should not
G. Valves, Piping, and Plumbing Details be exposed to sodium hydroxide sanitizing solution for long periods.
For lab and pilot scale equipment, plastic hose and tubing are adequate for I. Fraction Collectors
piping. Ball valves or pinch clamps serve as the valves to control flow. For
full production, stainless steel tubing and fittings that conform to the 3A Fraction collectors in the lab tend to be of the highly automated moving
sanitary code should be specified. Dead volume should be kept to a head type and thus easily adapted to the particular needs of the separation
at hand. As the scale of operation increases, the number of models of
minimum. Velocities through the tubing should correspond to mildly
turbulent flow if backmixing is to be kept to a minimum (55). Table 1, fraction collectors to choose from decreases and the range of options
adapted from Johansson (30), shows the size required for different flows with narrows. For production columns the fraction collector is often not a
aqueous solutions at room temperature. Sanitary diaphragm valves, butterfly separate unit. Rather it is a computer-controlled piping manifold. Once the
valves, or special ball valves free of inaccessible pockets serve for production program to control the valves in the manifold has been set, it is locked in.
It is possible to change it, but not as conveniently as with the pilot or lab The packing procedure for a large-scale column is identical to the lab
units. Another feature of production fraction collectors is that the number procedure except that the amount of gel added to the column has to be more
of fractions is generally much fewer than those taken in a lab separation. By carefully controlled. The fixed head means that the gel level has to be at
the time the separation has been scaled up, the number of fractions to take precisely the top of the column for efficient packing. Normally several
has been optimized. Each fraction from the large column goes to a separate attempts will be required before the level is just right. To load gel to the top
tank. of the column, a short extension piece must be clamped onto the top of the
column (31). The clamps should be of the type that can be snapped open
J. Packing the Column and removed in a few seconds. Welder's vise grips are suitable. Gel slurry
The column must be packed correctly if efficient chromatography is to be (75% v/v solids) is added to the extension piece in an amount estimated to
obtained. Although the specifics of packing vary depending on the type of bring the gel level to slightly above the joint between the extension piece and
gel and the manufacturer of the column, it is instructive to give two typical the top of the column. The extension piece is filled to the top with buffer.
procedures for the lab and large scale as examples of the precautions A rotary lobe pump is connected to the bottom drain of the column, and
necessary to achieve uniform packing. The lab procedure assumes a column buffer is pumped out of the column at a rate greater than that of the flow
with a movable head plate. The large-scale procedure assumes a fixed head at any step of the chromatography (16) until the buffer level is just above the
plate. gel. Exact recommendations for the flow rate are available from the
The lab procedure starts with filling the column with buffer until liquid adsorbent manufacturers. The pressure gradient caused by the pumping will
covers the lower bed support by about 2 inches. The bottom drain valve is compress the bed. If the gel estimate is correct, the gel level will be just
opened, and buffer is run out until all air is purged from the bottom piping. below the top of the column. The pump is stopped, and the clamps holding
Any air bubbles trapped under the bed support screen can be removed by the extension piece onto the column are quickly sprung and removed. As
vacuuming them out. This is done with a hose that sucks the bubbles the extension is removed, excess gel and buffer cascade onto the floor and
through the screen into a trap bottle connected to a vacuum source. The create quite a mess. The top head of the column is lowered very quickly
gel is suspended in enough degassed buffer to make the slurry about 75% onto the column. One must work quickly because the bed starts to expand
solids by volume (41). The buffer should match the composition of the as soon as the pumping stops. As soon as the lid is down, it is bolted in
process buffer that causes the greatest shrinkage of the gel. It has been place. Buffer is pumped upward through the column to displace air from the
found that packing a relatively heavy slurry gives a more uniform bed than top head and piping. The process required considerable skill and some luck.
packing a thin slurry. Care should be taken not to get air into the slurry The efficiency of the column can be checked by determining the HETP
either during stirring or upon addition to the column. The slurry is added as described by Cooney and others (16,69).
quickly to the column containing about 2 inches of free buffer. The best
results are obtained when the column is filled in a single pouring. VII. DEPYROGENATION AND COLUMN HYGIENE
Always keep a head of buffer over the gel surface. The last portion of A. Depyrogenation
gel should settle to a height within the adjustment of the top bed support.
Buffer is added to overflow the column. The vent on the top head plate is Pyrogens are substances that cause a fever response when injected into man
opened, and the top head is slightly tilted to the side before lowering it down or other animals. The most common variety are endotoxins derived from the
onto the column. The tilt prevents air from getting caught under the top bed cell wall of Gram negative bacteria such as E. coli. Endotoxins are high-
support, while the open vent allows air and water trapped in the head plate molecular-weight lipopolysaccharides of acidic character. Only tiny amounts,
plumbing to escape. Once the head plate makes a complete seal on the as little as 80 ng of E. coli pyrogen for the average man, are necessary to
column, it is straightened out and slowly lowered to the gel level. The head elicit the fever reaction (70).
plate is secured to the column, and final adjustments in height are made Virtually all protein products will contain pyrogens unless positive steps
before locking the support in place. The bed should have a uniform are taken to control them. Proteins derived from recombinant E. coli will be
appearance, free of trapped air bubbles. In making the connections to the heavily contaminated with pyrogens at the start of refining. Even in cases
column make sure that all air has been purged from the lines. Further fine where E. coli is not involved, pyrogens can still be expected originating from
adjustments may be required after the bed has been equilibrated with buffer. other natural raw materials (70).
The flow of buffer tends to consolidate the bed, producing a void at the Although pyrogens are derived from bacteria, they are not living.
entrance. Ordinary steam sterilization will not destroy them (71). Chemical destruction
Once the pyrogens have been removed, great care must be taken to
prevent recontamination. Pyrogens are ubiquitous. The most common
sources are water and textiles made from natural fiber. The buffer for the
elution step and all subsequent operations should be made up with water
that is pyrogen-free. Water for Injection, USP (WFI) is suitable (72). All
tanks and lines should be washed with WFI before introducing the eluted
protein solution (73). Paper towels and cotton wipes should not be used to
dry WFI-rinsed surfaces. Operators should wear surgical gloves for all
operations. The column itself can be depyrogenated by exposure to sodium
hydroxide for a length of time sufficient to hydrolyze the pyrogens. Figure
7 shows the time-concentration relationship for effective depyrogenation. A
typical set of conditions would be 0.5 N NaOH for 6 h at room temperature.
The hold period is important to allow time for complete chemical degrada-
tion.
B. Sanitization and Long-Term Storage

Columns must be sanitized between uses to prevent microorganisms from
colonizing the equipment. The adverse effects of such colonization include
breakdown of the column matrix, contamination of the product with
unknown bacteria and their by-products, buildup of pyrogens, possible
digestion of the product by proteases elaborated by the bacteria, and
plugging of the column. In most cases, sanitization, i.e., the near total
eradication of microbes, is sufficient to control the problem. Sterilization is
not usually necessary. Sanitization is best implemented by a clean-in-place
Figure 7 Destruction of E. coli endotoxin by sodium hydroxide at room tempera (CIP) procedure on a regular basis, preferable after every batch. With
ture. The effect of sodium hydroxide concentration and time. (Reprinted with properly designed equipment (sanitary design to 3A standards) it should be
permission from Pharmacia LKB Biotechnology.) unnecessary to unload the gel between uses. The most popular sanitizing
agent is sodium hydroxide. It possesses the following set of advantages as
first enumerated by Curling and Cooney (74) that put it far ahead of
requires extremes of temperature or pH that are far beyond the stability of
competing agents:
proteins.
How then to control and remove pyrogens? One of the most effective 1. Its presence in the final product is of no consequence.
ways of removing endotoxin is to pass the protein through an anion exchange 2. It is an effective disinfectant.
column. Most pyrogens are strongly acidic in nature (6). They bind to the 3. It has a solubilizing effect on proteins and lipids and will, therefore,
column even under conditions that would elute any adsorbed protein. The remove such residues that can act as substrates for bacterial growth.
purification factor can easily be 10,000 or greater. While the purification 4. It is inexpensive.
factor may be impressive, it just suffices in the face of the extreme potency 5. It is easily disposed.
of pyrogens. 6. It is compatible with the separation media used.
It is helpful to construct an endotoxin material balance across the total Other chemicals such as ethanol, isopropanol, sodium azide, chlorbuta-
separation process. The balance will be crude because the assay for nol, chlorhexidine, merthiolate, and bleach work perfectly well as disinfecting
endotoxin is quite variable. Even with this problem, it will be instantly clear agents, but they fail on one or more of the questions regarding toxicity,
where to concentrate pyrogen removal and control. Usually pyrogen levels reactivity, and carryover of residues to the final product. In special
will reduced by many orders of magnitude as the protein is purified. circumstances, such as resistance of the bacterial flora or instability of the gel
toward base, it may be necessary to adopt one of these compounds in place be validated by extensive testing with appropriate challenges. Every aspect
of sodium hydroxide. of the operation must be examined for possible sources of contamination.
It is good practice to monitor the microbiological condition of the Obviously the design, operation, and validation of an aseptic chromatog-
column on a regular basis. Later tests can be compared to the original work raphy system is an expensive and time-consuming task. Every effort should
to demonstrate the continuing effectiveness of the sanitization procedure. be made to postpone the sterility requirement to as late in the process as is
Resistant strains can develop over time or be introduced with new lots of raw consistent with product quality. The best solution, if possible, is heat
materials. If they are detected, switching to a different disinfectant or a sterilization of the final product (76). Unfortunately, most proteins would
much more concentrated solution of the same agent should clear up the be denatured by this treatment. More realistic alternatives are absolute
problem. Time and temperature of contact are important variables in filtration through a 0.22 micron filter or exposure to a sterilizing dose of
determining the effectiveness of the disinfection. radiation.
Long-term storage should be treated as a separate operation deserving
careful consideration. Once the column has been sanitized it can be kept in VIII. AUTOMATION
that condition by holding the gel in a weak solution of disinfectant. As an
example, sodium hydroxide at 0.05 N is commonly used for storage of ion A. Pilot Plant Requirements and Systems
exchange gels. Either the same or a different agent as that used for the main The distinguishing feature of the pilot plant is the need for flexibility (77).
sanitization may be chosen (75). The choice depends primarily on effective- This translates to the need for an automation system that can be adapted on
ness and lack of reactivity toward the gel over extended time. The dilute short notice with a minimum of programming skill. For serious development
disinfectant maintains a bacteriostatic action while minimizing degradation work, the control unit should be capable of duplicating the full chromato-
of the gel over long periods of time. If convenient, the column may be held graphic cycle anticipated in production. At the lab and pilot plant stage
at 4 °C to retard spoilage further. Certain gels can be completely dried for many automation systems are available that meet this requirement (78,79).
l ong-term storage (55). They are available off-the-shelf and often include a dedicated microprocessor
When it is time to use the gel again, it should be put through a full controller, pumps, and associated plumbing, all in one package. Most are
regeneration cycle to remove any degradation products that might have been quite flexible in adapting to the many different chromatographic separations
produced during storage. that are encountered in pilot plant work. Peak detection, fraction collection,
and step and gradient changes in eluent are easily handled. More trouble-
C. Sterilization some, but still doable, are reverse flow, switching from one step in the
In special circumstances it may be necessary to sterilize the adsorbent bed. chromatography to the next, and multiple buffers of very different composi
In these cases very careful design of the entire system is essential to be sure tion. The control unit can be separated from the rest of the system if it is
that all surfaces are contacted effectively by the sterilizing agent. The necessary to run the column in a refrigerated room. None of the systems on
preferred agent is live steam. Unfortunately, packed columns cannot stand the market are temperature programmable as would be useful for quicker
this type of treatment. Chemical sterilants such as ethanol, sodium and more effective simulations of the sanitization cycle.
hydroxide, or formaldehyde are possible substitutes. Sodium hypochlorite All of these units strive to be user-friendly. The best do succeed. The
and hydrogen peroxide should be avoided because they generate gas bubbles number of programs that can be stored is continually increasing with each
in the bed (30). The manufacturer of the adsorbent should be consulted for new model. There is also a trend toward having the ability to request a
specific recommendations regarding the type and concentration of sterilant complete operation with a single keystroke rather than having to key each of
necessary to sterilize their media without adversely affecting their adsorptive the individual valves that must be energized to accomplish the operation
properties. Naturally all piping, fittings, and solutions associated with the (79). Documentation of planned and ad hoc steps during the course of the
operation of the column must also be sterilized. All connections and separation is much better with recent machines. Paper tape indicating the
takeoffs must be capable of aseptic operation. position of the various valves has given way to a full video display of the
The Food and Drug Administration (FDA) has deservedly very strict ongoing chromatography in flowsheet form along with a complete history of
regulations covering aseptic operations (76). There are many technical the programmed and manual interventions. This reflects recognition by the
papers that cover the dos and don'ts of aseptic operation (29,30,75). These vendors of the importance of good documentation as proof of compliance
sources should be consulted before attempting to design an aseptic with FDA regulatory requirements. Some of the units can handle small
chromatographic separation. Whatever method is finally decided on, it must production jobs, such as the purification of monoclonal antibodies. The
Simpson Conventional Chromatography
250 251
upper limit of scale is set by the flow capacity of the valves, lines, and Automation also nicely handles the logistical problems of running multiple
controlled pumps. columns that are not synchronized in operation.
The alternative of a dedicated solid-state controller is cheaper for the The computer system itself should be validated following guidelines
hardware but decidedly user-unfriendly. At the other end of the spectrum, being developed by the FDA and the pharmaceutical industry (82). Proper
a multipurpose process computer does hold the advantage of being directly validation assures that the program will perform as predicted, that proper
translatable to production. The control algorithm can be developed at the backup and documentation is provided, that operators are properly qualified
pilot stage, and operating personnel will be familiar with it prior to plant and trained, and that the system is secure and reliable.
start-up. The disadvantages are the greater expense of the computer and the
time lost to the learning curve before operation is smooth. C. Current Good Manufacturing Practices
The final alternative is no automation at all. Depending on the The FDA has issued a set of regulations that are intended to insure the
turnaround time of the column, the frequency of production, and the production of safe and effective drugs. These regulations, entitled Current
complexity of operation, it is reasonable in many cases to operate columns Good Manufacturing Practices (CGMP) (83), cover the formulation and
manually. Most of these situations will arise at the pilot plant scale where packing of the bulk drug into a finished dosage form. Over the years the
supervision will be continuous with or without automation. agency has extended the published regulations back in the form of guidelines
(84) to cover the preparation of the bulk drug. In the case of drugs derived
B. Production Requirements and Systems from recombinant biotechnology, special points to consider (7) have been
Production automation must emphasize reliability, robustness, and a clearly issued.
documented history of each operation (77). The control of large-scale The approach in all cases is to foster a philosophy of building quality
chromatography steps is usually handled as part of a computer system that into every aspect of the process rather than just relying on the final quality
oversees the operation of the entire process. The types of computers and control assay to detect problems. This is done through proper equipment
software that are available for batch processing can quite easily handle the design, training, validated procedures, and documentation proving continuing
sequential steps of chromatography. Unlike the pilot plant models, the compliance. The significance of this with regard to scale-up of chromatogra-
piping, valves, and pumps for the production system are usually custom phy is that certain features must be designed into the system right from the
designed and interfaced with the controlling computer. Input and output start if compliance is to be assured. As the product gets closer and closer to
devices are designed for the higher loads and noisy electrical fields typically its final purity, it becomes more important to protect that purity from
found in industrial operation. contamination. Many times, chromatography steps are near the end of the
The control language for plant systems is not as customized as that for process and represent the final purification. Subsequent steps just serve to
pilot plant systems. The burden of dealing with a slightly less friendly system concentrate or dialyze the product. In these circumstances, the design should
is compensated for by universal applicability once it is mastered. Production protect against the introduction of foreign matter and cross-contamination
control programs are protected from tampering by passwords and key locks. with other products handled in the plant. The equipment must be designed
While it does protect the integrity of the program, it also makes it more so that it can be easily cleaned, and the gel must be reserved for one product
cumbersome to make a legitimate change. This is a feature common to all only (19). The chromatography hardware must be dedicated to a single
production systems. It is assumed that the control strategy has been well product or be cleaned by a method shown to be effective by actual test.
worked out before the protein is put into production. Changes occur much Water for the chromatography must be of the appropriate quality, in
less frequently than in a pilot plant. most cases WFI. A testing program must be in place to monitor the water
Accurate record keeping is one of the major advantages of automation. quality (85), the pyrogen content of the eluate, and the bioburden of the
All production control systems must provide a good paper trail of all actions buffers, column, and eluate. Closed systems are preferred over open tanks.
taken by the computer and operators. Chromatography in production should In most situations, rated cleanrooms (86) are appropriate for the final steps
have already undergone an extensive validation study (80,81) to show that the of the process.
purification is effective when certain operating parameters are maintained Every aspect of the operation and control of the chromatography must
within prescribed limits. A major aim of automatic control is to prevent be properly documented with procedures and batch records. All raw
deviations outside of these limits. In cases where the system does go out of materials must be covered by an inventory control system designed to insure
control, the deviation is documented and highlighted for corrective action. their identity and potency.
252
APPENDIX: LIST OF SYMBOLS Subscripts

E elution phase
H holding phase
L loading phase
R regeneration phase
W washing phase
Greek
REFERENCES
1. Sofer, G. K., and Nystrom, L. E. (1989). Process Chromatography, A Practical
Guide, Academic Press, London, pp. 107-115, 65, 93-105.
2. Scopes, R. K. (1987). Protein Purification Principles and Practice, 2nd ed.,
Springer-Verlag, New York, pp. 100-126, 133, 161, 186-198.
3. Gooding, K. M., and Schmuck, M. N. (1983). Purification of trypsin and other
basic proteins by high-performance cation-exchange chromatography, J. Chroma-
togr., 266: 633.
4. Regnier, F. E. (1984). High-performance ion-exchange chromatography,
Methods in Enzymology, 104: 170.
5. Richey, J. S. (1984). Optimal pH conditions for ion exchangers on macroporous
supports, Methods in Enzymology, 104: 223.
6. Sofer, G. (1984). Chromatographic removal of pyrogens, Bio/Technology, 2:
1037.
7. Points to Consider in the Production and Testing of New Drugs and Biologicals
Produced by Recombinant DNA Technology (1983). Draft, U.S. Department of
Health and Human Services, Public Health Service, Food and Drug Administra-
tion, Bethesda, Maryland, November 7.
8. Delaney, R. A. M. (1980). Industrial gel filtration of proteins, Applied Protein
Chemistry (R. A. Grant, ed.), Applied Science, Barking, England, pp. 233-280.
9. Gerstenberg, H., Sittig, W., and Zepf, K. (1980). Processing of fermentation
products, Ger. Chem. Eng., 3: 313.
10. Kelley, J. J., Wang, G. Y., and Wang, H. Y. (1986). Large-scale gel chromatog-
raphy: Assessment of utility for purification of protein products from microbial
Simpson Conventional Chrontatography 255
254
27.
sources, Separation, Recovery, and Purification in Biotechnology, A. C. S. Chen, J. W., Buege, J. A., Cunningham, F. L., and Northam, J. I. (1968). Scale-
Symposium Series, No. 314 (J. A. Asenjo and J. Hong, eds.), American Chemical up of a column adsorption process by computer simulation, Ind. Eng. Chem.
Society, Washington, D.C., pp. 193-207. Process Design and Development, 7: 26.
Hammond, P. M.., Sherwood, R. F., Atkinson, T., and Scawcn, M. D. (1987). 28. Cueille, G., and Tayot, J. L. (1985). "Ion Exchange as a Production Technique
11.
Hydrophobic interaction chromatography for large scale purification of proteins, for Proteins," Proceedings of Biotech '85 Europe, Geneva, pp. 141-160.
29. Simpson, J. M. (1988). Preliminary design of ion exchange columns for protein
Chim. Oggi, 10: 57.
Ingraham, R. H., Lau, S. Y. M., Taneja, A. K., and Hodges, R. S. (1985). purification, Bio/Technology, 6: 1158.
12. 30.
Denaturation and the effects of temperature on hydrophobic-interaction and Johansson, H., Oesting, M., Sofer, G., Wahlstroem, H., and Low, D. (1988).
reversed-phase high performance liquid chromatography of proteins. Bio-Gel Chromatographic equipment for large-scale protein and peptide purification,
TSK-Phenyl-5-PW column, J. Chromatogr., 327: 77. Adv. Biotechnol Processes, 8: 127.
Groundwater, E. (1985). Guidelines for chromatography scale-up, Lab. Pract., 31. Janson, J. C., and Dunhill, P. (1974). "Factors Affecting Scale-Up of Chroma-
13.
34: 17. tography," Proceedings of FEBS Meet., 30, pp. 81-105.
32.
14. Sluyterman, L. A. A., and Wijdenes, J. (1981). Chromatofocusing IV. Colin, A., Lowy, G., and Cazes, J. (1986). Design and performance of prepara-
Properties of an agarose polyethyleneimine ion exchanger and its suitability for tive-scale HPLC, International Biotechnol. Laboratory, 4:30.
33. Arnold, F. H., Blanch, H. W., and Wilke, C. R. (1985). Analysis of affinity
protein separations, J. Chromatogr, 206: 441.
Bergstrom. J., Sodenberg, L., Wahlstrom, L., Muller, R. M., Momicelj, A., separations I: Predicting the performance of affinity adsorbers, Chemical
15.
Hagstrom, G., Stahlberg, R., Kallman, I., and Hansson, K. A. (1985). Recom- Engineering J. , 30: B9.
mendation for obtaining optimal separations on MONOBEAD" ion exchangers, 34. Arnold, F. H., Blanch, H. W., and Wilke, C. R. (1985). Analysis of affinity
Protides of the Biological Fluids (H. Pecters, ed.), Pergamon Press, Oxford, vol. separations II: The characterization of affinity columns by pulse techniques,
30, pp. 641-646. Chemical Engineering J., 30: B25.
Cooney, J. M. (1984). Chromatographic gel media for large-scale protein 35. Arnold, F. H., Schofield, S. A., and Blanch, H. W. (1986). Analytical affinity
16.
chromatography I: Local equilibrium theory and measurement of association
purification, Bio/Technology, 2: 41.
Kroeff, E. P., Owens, R. A., Campbell, E. L., Johnson, R. D., and Marks, H. 1. and inhibition constants, J. of Chromatogr., 355: 1.
17.
(1989). Production scale purification of biosynthetic human insulin by reversed- 36. Arnold, F. H., and Blanch, H. W. (1986). Analytical affinity chromatography II:
phase high-performance liquid chromatography, J. Chromatogr., 461: 45. Rate, theory, and measurement of biological binding kinetics, J. of Chromatogr.,
Sadana, A. (1989). Protein inactivation during downstream processing, Part I: 355: 13.
18.
The processes, BioPharm., 2: 14. 37. Cowan, G. H., Gosling, I. S., Laws, J. F., and Sweetenham, W. P. (1986).
19. Duffy, S. A., Moellering , B. J., Prior, G. M., Doyle, K. R., and Prior, C. P. Physical and mathematical modelling to aid scale-up of liquid chromatography,
(1989). Recovery of therapeutic-grade antibodies: Protein A and ion-exchange J. Chromatogr., 363: 37.
chromatography, BioPharm, 2: 34. 38. Foo, S. C., and Rice, R. G., (1975). Prediction of the ultimate separation in
parameterc pumps, AIChE J., 21: 1149.
20. Chase, H. A. (1984). Scale-up of immunoaffinity separation process, J. of
Biotechnol, 1: 67. 39. Young, M. E., Carroad, P A., and Bell, R. L. (1980). Estimation of diffusion
21. Frenz, J., and Horvath, C. (1988). High-performance displacement chromatogra- coefficients of proteins, Biotechnol Bioeng, 22: 947.
40. Miller, S. F., and King, C. J. (1966). Axial dispersion in liquid flow through
phy, High-Performance Liquid Chromatography, vol. 5 (C. Horvath, ed.),
Academic Press, San Diego, California, pp. 212-314. packed beds, AIChE J., 12: 767.
Katti, A., and Guiochon, G. (1988). Prediction of band profiles in displacement 41. Pharmacacia LKB Biotechnology, Piscataway, N. J. (1989). "Preparative Gel
22.
chromatography by numerical integration of a semi-ideal model, J. Chromatogr, Filtration Goes High Performance," Downstream: News and Views for Process
449: 25. Biotechnologists, 7.
42. Pharmacia Fine Chemicals AB, Sweden (1979). Gel Filtration- Theory and
23. Subramanian, G., Phillips, M. W., Jayaraman, G., Cramer, S. M. (1989).
Displacement chromatography of biomolecules with large particle diameter Practice, p. 42.
systems, J. Chromatogr., 484: 225. 43. Kopaciewicz, W., and Regnier, F. E., (1983). Nonideal size-exclusion chromatog-
24. Felinger, A., and Guiochon, G. (1993). Comparison of maximum production raphy of proteins: Effects of pH at low ionic strength, High-Performance Liquid
rates and optimum operating/design parameters in overloaded elution and Chromatography of Proteins and Peptides, (M.T.W. Hearn, F. E. Regnier, and C.
displacement chromatography, Biotechnol. Bioeng., 41: 134. T. Wehr, eds.), Academic Press, New York, pp. 151-159.
25. Curling, J. (1979). "Large-Scale Ion Exchange Chromatography of Proteins," 44. Afeyan, N. B., Fulton, S. P., Gordon, N. F., Mazsaroff, I., Varady, L., and
Proceedings of FEBS Meet., 52, pp. 45-57. Regnier, F. E. (1990). Perfusion chromatography, an approach to purifying
Van Brunt, J. (1988). Scale-up: How big is big enough, Bio/Technology, 6: 479. biomolecules, Bio/Technology, 8: 203.
26.
45. Van Deemter, J. J., Zuidenveg, F. J., and Klinkenberg, A. (1956). Longitudinal 64. Keuer, T. A., and Ryland, J. R. (1990). "Industrial-Scale Ion Exchange
diffusion and resistance to mass transfer as causes of nonideality in chromatogra- Chromatography for Purification of Proteins," Abstract from Engineering
phy, Chem. Eng. Sci., 5: 271. Foundation Conference: Recovery of Biological Products V, St. Petersburg,
46. Creighton, T. E. (1984). Proteins, W. H. Freeman, New York, p. 242. Florida, May 13-18.
47. Pfannkoch, E., Lu, K. C., Regnier, F. E., and Barth, H. G. (1980). Characteriza- 65. Amicon Division, W. R. Grace and, Danvers, Mass. (1988). "Stainless Steel
tion of some commercial high performance size exclusion chromatography Columns for Pilot and Process Scale Chromatography," Pub. 855.
columns for water soluble polymers, J. Chromatogr. Sci., 18: 430. 66. Janson, J. C. (1971). Columns for large-scale gel filtration on porous gels, J. Agr.
48. Eveleigh, J. W., and Levy, D. E. (1977). Immunochemical characteristics and Food Chem., 19:581.
preparative application of agarose-based immunosorbent,J. Solid-Phase Biochem, 67. Janson, J. C. (1978). "Large Scale Chromatography of Proteins," Proceedings
2: 45. Int. Workshop Technol. Protein Sep. Improv. Blood Plasma Fractionation, 1977,
49. Kunii, D., and Levenspiel, O. (1969). Fluidization Engineering, John Wiley, New DHEW Publ. (NIH) (U.S.), NIH-78-1422, pp. 205-220.
York, p. 67. 68. Scopes, R. K. (1974). Measurement of protein by spectrophoto metry at 205 nm,
50. Fulton, S. P. (1989). TheArt ofAntibody Purification, Pub. 868, Amicon Division, Analytical Biochemistry, 59: 277.
W. R. Grace, Danvers, Mass., p. 57. 69. Pharmacia LKB Biotechnology, Piscataway, N.J. (1989). "Get the Best from your
51. Holdoway, M. J. (1989). The design of silica for HPLC, Industrial Chromatogra- Sephaeryl HR," Separation News, 16.1: 5-8.
phy News, 1: 8. 70. Pearson, F. C. (1985). Pyrogens: Endotoxins, LAL Testing and Depyrogenation,
52. Industrial-Scale Chromatographic Separation with High Performance Dowex Marcel Dekker, New York, pp. 11, 23, 37, 24.
Monosphere Resins, Form No. 177-1371-84, Dow Chemical USA, Specialty 71. Novitsky, T. J., and Gould, M. J. (1985). Depyrogenation by moist heat,
Chemicals Dept. Depyrogenation, Technical Report No. 7, Parenteral Drug Association Inc.,
53. Yamamoto, S., Nakanishi, K., and Matsuno, R. (1988). Ion Exchange Chroma- Philadelphia, Penn., p. 109.
tography of Proteins, Chromatographic Science Series No. 43, Marcel Dekker, 72. United States Pharmacopeial Convention, Rockville, Md. (1990). United States
New York, p. 306. Pharmacopeia, p. 1456.
54. Scawen, M. D., and Hammond, P. M. (1989). Fractionation techniques in 73. Weary, M. (1985). Depyrogenation, Pyrogens: Endotoxins, LAL Testing and
process biotechnology, J. Chem. Tech. Biotechnol., 46: 85. Depyrogenation (F. C. Pearson, ed.), Marcel Dekker, New York, p. 209.
55. Janson, J. C., and Hedman, P. (1982). Large-scale chromatography of proteins, 74. Curling, J. M., and Cooney, J. M. (1982). Operation of large-scale gel filtration
Advances in Biochemical Engineering, vol. 25, Chromatography (A. Fiechter, ed.), and ion-exchange systems, J. Parenteral Science and Technology, 36: 59.
Springer-Verlag, Berlin, pp. 43-99. 75. Hasko, F., Kristof, K., Salamon, M., and Dobo, P. (1985). Large scale chromato-
56. Satterfield, C. N., Colton, C. K., and Pitcher, W. H. (1973). Restriction diffusion graphic experiments for plasma, Dev. Hematol. Immul, 13: 105.
in liquids within fine pores, AIChE J., 19: 628. 76. U. S. Department of Health and Human Services, Public Health Services, Food
57. Regnier, F. E. (1982). Review: High performance ion-exchange chromatography and Drug Administration (1987). Guideline on Sterile Drug Products Produced
of proteins: The current status, High Performance Liquid Chromatography of by Aseptic Processing, June.
Proteins and Peptides (M. T. W. Hearn, F. E. Regnier, and C. T. Wehr, eds.), 77. Johansson, J. (1985). "Automation of Chromatographic Processes," Proceedings
Academic Press, New York, pp. 1-7. of Biotech '85 Europe, Geneva, May, 1985, pp. 193-202.
58. Pharmacia LKB Biotechnology, Piscataway, N.J. (1986). "Process Hygiene in 78. Pharmacia LKB Biotechnology, Piscataway, N.J. (1988). FPLC for Protein
Industrial Chromatographic Processes," Downstream: News and Hews for Process Chromatography, Pub. No. FFPC 50-01-425.
Biotechnologists, Form 60-01-019 PD02. 79. OROS Instruments, Cambridge, Mass. (1989). OROS Instruments Controlled
59. Viljoen, M., Shapiro, M., Crookes, R., Chalmers, A., and Marrs, S. (1982). Protein Liquid Chromatography.
"Large-Scale Recovery of Human Serum Albumin by a Chromatographic 80. U.S. Department of Health and Human Services, Public Health Services, Food
Method," Abstract Th-131 of the Joint Meeting of the 19th Congress of the and Drug Administration, (1987). Guidelines on General Principles of Process
International Society of Haematology and the 17th Congress of the International Validation, May.
Society of Blood Transfusion, Budapest, Aug. 1-7. 81. Jones, A. J. S., and O'Connor, J. V. (1985). Control of recombinant DNA
60. Knight, P. (1989). Chromatography: 1989 report, Bio/Technology, 1: 243. produced pharmaceuticals by a combination of process validation and final
61. Amicon Division, W. R. Grace, Danvers, Mass. (1988). "Moduline Chromatogra- product specifications, Develop. Biol. Standard., 59: 175.
phy Columns for Pilot and Process Scale Chromatography," Pub. 852. 82. Agalloco, J. (1989). Validation of existing computer systems, Pharmaceutical
62. Amicon Division, W. R. Grace, Danvers, Mass. (1988). "Acrylic Columns for Technology, Jan: 38.
Pilot and Process Scale Chromatography," Pub. 853. 83. Human and Veterinary Drugs Good Manufacturing Practices and Proposed
63. Amicon Division, W. R. Grace, Danvers, Mass. (1988). "Glass Columns for Pilot Exemptions for Certain OTC Products (1978). Federal Register, Part 11, Sept. 29,
and Process Scale Chromatography," Pub. 854. 1978, 43 (190): 45013-45089.
258 Simpson
84. U.S. Department of Health and Human Set-vices, Public Health Service, Food
and Drug Administration (1991). Guide to Inspection of Bulk Pharmaceutical
Chemicals, Washington, D. C., September.
85. Pharmaceutical Manufacturers Association Deionized Water Committee (1985).
Validation and control concepts for water treatment systems, Pharmaceutical
Technology, Nov: 50.
86. Munson, T. E. (1988). "Regulatory View of Environmental Monitoring,"
Reprint from Pharm. Tech. Conference of September 1988.
Biospecific Affinity Chromatography
Nikos K. Harakas
Monsanto Company, St. Louis, Missouri
I. INTRODUCTION
The purpose of this chapter is to describe succinctly for the practicing
chemical, biochemical, and process engineers and research and development
scientists and investigators the powerfulness of biospecific affinity chromato-
graphy (BAC) to obtain highly purified proteins and enzymes that can be
used for either in vivo or in vitro applications. BAC was championed in the
late 1960s by Cuatrecasas, Wilchek, and Anfinsen (1). They demonstrated
that enzymes could be purified using a specific competitor inhibitor
i mmobilized on solid matrices covalently attached. This development
followed the pioneering work in the early 1950s by Campbell, Leuscher and
Lerman (2), who isolated nonprecipitating antibodies from immune serum
by means of a cellulose-protein antigen. Later, in 1953, Lerman (3) extended
the crude affinity technique for the isolation of the enzyme tyrosinase using
azo dyes prepared by coupling an aromatic amine with phenol on cellulose,
which inhibits the enzyme.
Since the paper of Cuatrecasas and collaborators, over 1000 proteins and
enzymes have been purified by affinity chromatography, and probably over
10,000 papers have elaborated on the technique with major improvements.
There have been three striking developments in affinity chromatography in
the last two decades. First of these is the use of heavy metal ions such as
Zn2+ and Cue' as affinity ligands immobolized appropriately as chelates on
solid gels that bind selectively on histidine and cysteine residues on the
surface of proteins, successfully demonstrated in 1975 by Porath and collabo-
rators (4). Second is the use of monoclonal antibodies as ligands, first dem-
259

Protein Purification Process Engineering

Uploaded by

Copyright:

Available Formats

Protein Purification Process Engineering

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Protein Purification Process Engineering

Uploaded by

Copyright:

Available Formats

BIOPROCESS TECHNOLOGY

1. Membrane Separations in Biotechnology, edited by W. Courtney

ADDITIONAL VOLUMES IN PREPARATION

Recombinant Microbes for Industrial and Agricultural Applications,

ACS Symposium Series 427 (M. R. Ladisch, R. C. Willson, C. C.,

well analyzed in the literature. Regeneration, elution, and washing, however,

C. Formulation of the Design Problem

Bw B E , and B R are constants specific to each process. If the wash step is a

4. Final Design Equation

At low feed potency the number of columns is independent of bed height

volume. The highest-molecular-weight components elute no sooner than the

D. Determining the Rate Controlling Step

If the particles are extremely small or the protein-adsorbent interaction

(27) E. Effect of Process Variables

Alternatively, the manufacturer's tests of protein exclusion can be consulted

Figure 6 Amicon design of flow distributor for chromatography column.

operation of large industrial columns (64). Adsorption of seemingly colorless

B. Sanitization and Long-Term Storage

APPENDIX: LIST OF SYMBOLS Subscripts

Biospecific Affinity Chromatography

You might also like